THE UNIVERSE’S SYSTEMATIC ERROR

Chapter 1. Systematic Error

There are two common ways of thinking about who we are and what consciousness is. And both, if you look closely, say the same thing.

The first is that we are a unique creation. In the midst of cold and largely empty space, on a tiny planet at the edge of an otherwise unremarkable galaxy, something so incredible arose that it cannot be a coincidence: a mind capable of gazing at the stars and asking where it came from. And since it cannot be a coincidence—that means someone or something planned it.

The second view is the loneliest: we are a happy accident in an indifferent universe, where no one is waiting for us and no one will notice when we are gone. Just a lucky combination—a roll of the dice, molecules that happened to come together just right—and here we are: for a short time, without reason, without a future.

In both scenarios, consciousness often turns out to be something exceptional: either a miracle or an incredible win in the cosmic lottery. These two scenarios have been at odds for so long and so fiercely that it’s easy to overlook the main point: they share a common foundation. Both design and chance rest on a single unspoken assumption—that consciousness is an anomaly. Something that falls outside the normal course of events, outside physics itself, and therefore requires a special explanation: either a miracle or incredible luck without cause or effect. The debate centers solely on which of the two explanations to choose. No one disputes that, in essence, there is nothing to explain. This discussion has long resembled running in circles, with the search for an answer left far behind.

I propose a third way.

We are neither a design nor a coincidence. We are a systematic error. The result of a process that has no goal but does have logic. Not a miracle—because no miracle was required. Not a coincidence—because where the conditions came together, it was bound to happen. An anomaly—because it is a rare, extreme deviation, not the norm. And a natural one—because it is woven into the very fabric of the physical world, anticipated by it as calmly as rust is anticipated on iron left out in the rain.

To see this logic, I suggest we embark on a great journey—and start from afar. Not with a human being. Not with the brain. And not even with life. We must start at the very beginning—with what seems to be the complete opposite of any order and any meaning. With entropy.

* * *

Milk dissolves in coffee, the hot cools, the new decays, the living dies. The universe, if we are to believe the popular narrative, is a slow and inevitable descent into disorder, into a gray, warm mush where nothing else happens. This end even has a name, both beautiful and terrifying: heat death.

All of this is true. But it’s only half the truth.

Here’s what people usually leave out. The law that drives the world toward disorder—the second law of thermodynamics—is formulated for isolated systems. Sealed-off systems. Systems through which nothing flows: neither energy nor matter. A thermos forgotten in a vacuum. But almost nothing in the world around us is sealed off. A flow passes through everything. The Sun’s light flows through the Earth. Food flows through living things. Wax and air flow through a candle’s flame. We are not sealed-off thermoses—we are channels through which something flows.

In open systems, the second law of thermodynamics does not preclude local order. Moreover, with a sufficient flow of energy, order sometimes turns out to be the way in which the system most efficiently dissipates that flow.

Take a pot of water and place it over low heat. At first, nothing happens—the water is just water. But turn up the heat, create a sufficient temperature difference between the bottom and the surface—and at some point, the water ceases to be just water. A tremor runs through it of its own accord, and the chaotic jostling of molecules suddenly forms into neat cells: tiny rotating shafts, even hexagonal honeycombs, as if someone had drawn lines through the liquid with a ruler. No one drew them. No one commanded the molecules to arrange themselves. They arranged themselves—because it’s more efficient that way. But more efficient for what?

Here’s the twist that started it all, and it turns our familiar picture upside down. These patterns arise not in defiance of the second law of thermodynamics, but thanks to it. A well-coordinated structure conducts heat through itself faster and dissipates energy more readily than a chaotic jumble of molecules. Order here is not a rebellion against decay. Order here is a way to decay more efficiently. The universe builds structure not because it loves beauty, but because structure helps it do exactly what it strives to do: dissipate, waste, and approach equilibrium. Order turns out to be not the enemy of entropy, but its best worker.

Such self-assembling patterns in the flows passing through them are called dissipative structures—from the Latin dissipare, meaning “to disperse.” Ilya Prigogine described them and received the Nobel Prize for doing so, because his description, for all its modesty, quietly changed a great deal.

* * *

Take a moment sometime to watch a candle flame for a while. It hangs above the wick, flickering, yet it holds its shape—that same, recognizable drop of light, hour after hour. And yet nothing in it is constant: the wax rises, evaporates, burns, and drifts away as smoke and heat. At every moment, the flame consists of different matter. It is not the matter that is constant—it is the form. A flame is not a thing; it is a process that sustains itself as long as fuel flows through it. This is a dissipative structure, the most vivid example of all.

And as soon as you see this, you’ll begin to recognize it everywhere. A whirlpool in a stream that stands still, even though water flows through it without stopping. A hurricane that survives for weeks, sustained by the warm ocean beneath it. A plume of smoke that maintains a slender shape until it breaks into curls. All of these are islands of stubborn form in a river of decay, and they all persist for one reason: a current flows through them, and they have learned to disperse it. And at the very end of this line, immeasurably more complex, stand we. We, too, are a flame. Just a very slow one, and a very cleverly constructed one.

* * *

But it’s not that simple. It’s tempting to declare: since complexity helps disperse energy, then the universe must become more complex; complexity is embedded in the laws of nature as a command. There’s even a hypothesis along these lines—that non-equilibrium systems strive to produce entropy as quickly as possible and therefore naturally gravitate toward complex forms. But this is still a hypothesis, not a proven law, and it has both supporters and serious opponents. I will not mistake wishful thinking for reality, but will trust only what has been tested by time, logic, and common sense.

In the presence of an energy flow and an open system, increased complexity is not guaranteed at every single point. But it is predictably expected—statistically, on average, as a trend. The universe does not choose complexity nor does it strive for it. It simply becomes more complex wherever physically possible—for the same simple reason that water finds any crack: not out of inspiration, but because what is possible given time and energy will happen sooner or later. And the universe had an abundance of time and energy.

And if this is so—if complexity is neither a miracle nor a rare stroke of luck, but rather something that flows naturally from physics itself, wherever it finds a channel—then we will have to reexamine everything that stands at the pinnacle of complexity. Including that which gazes upon this pinnacle from within.

This is where we’ll begin our ascent. We’ve essentially already seen the first step: the pattern in a pot, a flame, a whirlpool—matter that has learned to hold its shape in a flow. There will be five more steps, and they’ll take us far—toward the living, toward the sentient, toward the one who asks questions. But no matter what changes from step to step, the same movement will repeat itself on each one—the same four-beat rhythm. Hear it once—and you’ll recognize it again and again, higher and higher, until, at the very top, this rhythm coalesces into the one who is capable of hearing it.

Chapter 2. The Circle

The flame holds its shape, but does not create itself. It burns on someone else’s fuel, lit by someone else’s hand; blow out the draft—and it will vanish in an instant. A flame cannot be wounded and cannot be healed: it has no parts to repair, no reserve from which to rebuild itself. It is a pattern, not the master of the pattern. And in this it is doomed: when the flow continues, it exists; when it breaks off, it ceases to exist, and between these two states there is nothing that could be called an effort to survive.

How, then, does a pattern become the master? Not by a sudden leap, nor by a miracle—but by trial and error.

The universe is generous with trials. It generates countless dissipative structures, especially in chemistry, where streams of matter weave together into thousands of fleeting patterns. And sooner or later, a special one appears among them: it doesn’t just dissipate energy—it simultaneously accelerates the reaction that produces its own component. One substance, as it arises, helps another to arise.

Most often, such a chain leads nowhere: one helps the second, the second helps the third, the third dissipates—and everything fizzles out. But occasionally—purely by chance, since there are so many possibilities—the chain forms a loop: the last link helps the first one come into being. And then something happens that has never occurred in any structure before : the loop begins to sustain itself. The products of its work become its own raw materials.

* * *

Here it is, the fourth beat of the rhythm—the leap upward. As long as the chains remained open, the structure was a pattern in the flow, like a flame. Once the ring closed—the pattern became the master: it now produces the parts of which it is composed.

And such a ring cannot be washed away—not because someone chose or preserved it: there is no one in the universe to choose. But because a ring that rebuilds itself faster than it falls apart lasts longer—and everything that does not rebuild itself is washed away by the flow. The universe does not select self-assembling structures. It loses all the rest—and at the bottom, when the foam of random attempts settles, what remains is that which can hold itself together.

This self-sustaining system has a name: autopoiesis, or self-creation. It was coined by biologists Humberto Maturana and Francisco Varela to describe what distinguishes the living from the non-living—not by substance, but by structure: the living is not a collection of specific molecules, but a closed loop in which the system continuously produces itself. The simplest example of this is a living cell. A bag of chemistry that is both the bag and the chemistry itself.

And along with the loop, a boundary appears in the world for the first time—a real one, drawn from within. A flame has no boundary: it is wherever the flow occurs, and we draw its edge from the outside. A cell has its own boundary—it maintains, repairs, and protects it. For the first time, a fragment of the universe draws a line around itself and, through chemistry alone—without words or thought—establishes a distinction upon which everything else will later be built: this is me, and this is not me.

It is still an infinitely long way from here to consciousness. But the first stone in its foundation has already been laid—the boundary between the self and the world.

* * *

There is another acquisition, imperceptible but decisive. As long as the structure lived off an external flow, it had no time of its own: it ticked in time with what flowed through it. The ring, however, repairs itself faster than it disintegrates—and in doing so, detaches its own rhythm from that of its environment. It starts its own clock: not an astronomical one, but an internal one—the pace at which it manages to reassemble itself before it falls apart. From this first clock begins everything that will later become memory, anticipation, and, much later, the experience of duration.

We have completed the first full cycle of the rhythm and can now hear it in its entirety. The distinction upon which the system rests is a gradient: for a cell, this is a chemical difference across the boundary. The way to avoid disintegration is retention: a closed loop of self-repair. The thing around which everything is built is the pole: for now, it’s simply the stability of the structure itself, and you can almost ignore it; but remember it, because this third beat will someday become the main character of our story. And the leap upward is the transition: the closing of the loop we just saw. Hear this rhythm once—and from then on, you’ll recognize it at every step, growing ever more complex and ever more incredible.

So, the universe has drawn its first line and set its first clock in motion. It now has an “I”—tiny, chemical, wordless, but its own: it knows how to hold itself together and distinguish itself from everything else. What it cannot yet do is look ahead. It lives entirely in the “now”: if there’s a gradient, I hold on; if not, I disintegrate—and not a second sooner. The next stage will begin the moment this “self” is guided not by the reality of the present, but by a blueprint of what does not yet exist.

Chapter 3. The Blueprint

The Ring lives entirely in the “now.” If there’s a gradient, it holds; if not, it falls apart; it responds to what has already happened and is always one step behind. The world gave it a push—it swayed. As long as the world is kind, that’s enough. But the world is rarely kind twice in a row, and whoever merely reacts to a blow will sooner or later be too late to respond.

The difference is clear in a simple example. A standard thermostat waits until the room gets cold, and only then turns on the heat: it responds to what has already happened. Now imagine a thermostat that’s a little smarter—it knows it always gets colder at night and turns on the heat in advance, before you even start to feel cold. The first reacts to the present. The second acts according to a blueprint of the future. The next step is all about this transition from the first to the second.

In living organisms, it looks like this. Imagine two cells in a drop of water where food is becoming increasingly scarce. The first waits until the food touches its wall, and only then reaches out for it. The second is structured slightly differently: it has a sort of chemical compass inside that, following the fading trail of nutrients, turns it toward where there is more food—even before any contact is made. Which one will survive longer in water where food is running out? The answer is obvious, and in it lies the entire next stage.

The second cell does something unprecedented: it acts not based on what is, but on what is not yet there. Inside it is a tiny map of the world: “Food will come from where the smell is stronger.” And it moves according to this map, staying one step ahead of events.

And once again, the universe doesn’t choose those who foresee—it simply discards the rest. In scarce water, a cell with an internal compass adapts better than one that waits for contact. Foresight is neither a luxury nor a trick, but a way to survive better; and so, once it arises, it takes hold, just as the ring took hold before.

But where does the cell get this blueprint? It doesn’t fall from the sky, nor does it switch on all at once. It grows out of what the ring already possesses—and that’s the whole point: no magic, just an assembly of ready-made parts.

Part One—Reflection. The autopoietic ring already possesses internal states, and these states are already changing in tandem with the world. If conditions outside become harsher, the internal equilibrium shifts; if it craves food, its chemistry runs a little differently. This isn’t a model yet; it’s a simple interdependence: the internal involuntarily mirrors the external because it’s connected to it. But it is precisely this involuntary mirroring that serves as the raw material from which the blueprint will emerge.

Part Two—memory. We’ve already seen that a cell lives at its own rhythm and therefore involuntarily retains a trace of what just happened to it. And the retained trace of yesterday is already a tiny prediction of tomorrow: “it smelled like food here” means “there will most likely be food here again.” Memory, turned forward, is the simplest form of prediction.

Part Three—consolidation, which brings the first two parts together. Among the cells whose internal state already reflects the world and contains a trace, one occasionally encounters a cell in which this internal state is additionally linked to an action: the balance shifted toward “food is there”—and the flagellum beat precisely in that direction. Such a cell survives better—and becomes established. This is where the internal state ceases to be a passive mirror: it begins to replace the world itself. The cell no longer acts based on the scent itself, but on its internal “food is over there.” This substitution—when the internal takes the place of the external and governs behavior—is the birth of a model. Not a flash, but a fusion of three pre-existing elements: an involuntary reflection, a retained trace, and the consolidation that links them to action.

* * *

There’s one thing I need to clarify right away, or else it will lead us astray. When I call this a “model,” I don’t mean a mental image, a conscious representation, or a symbol. Here, “model” is used in its weakest, most practical sense: an internal state that is consistently correlated with the external world and capable of guiding action instead of direct contact with the object. Not an image of food, but an internal marker—“there is food”—that one can follow without waiting for the food itself. In this modest sense—and only in it—does the cell acquire a model. Anything richer—an image, a symbol, a thought—will come much later and will be built upon this modest beginning.

There is also a deeper aspect to this, one worth pausing to consider for a moment. In order to resist disintegration at all, a system is forced to behave as if it knows its world in advance: a system that cannot distinguish between what is safe and what is poisonous will collapse at the very first shift in its environment. To maintain itself within a certain world already means, in a hidden way, to carry its model. Carl Friston demonstrated this rigorously: a system that stubbornly resists dispersion is mathematically indistinguishable from a system that predicts its own niche. The model is not attached to life from the outside—it is the inner workings of self-sustainment. As soon as the ring began to take itself seriously, the model was already smoldering within; all that remained for our hero was to fan the flames and set it in motion.

It was he, Friston, who gave a name to what a living being does with the model next—active inference. The system tirelessly compares the blueprint within and reality without, and strives to close any gap between them: either by adjusting the blueprint to fit the world—in which case it learns—or by adjusting the world to fit the blueprint—in which case it acts. A hungry cell, drawn by scent, is engaged in the latter: it does not wait for the world, but brings the world into alignment with the blueprint, where it is already near the food.

Here is a new gradient—of a different kind than the previous ones. Before, the gradient was in space (heat below, cold above) or in chemistry (dense inside, empty outside). Now the gradient is in time: between what the model predicts and what actually exists. For the first time, the system distinguishes not between “here and there,” but between “now and later.” And for the first time, it truly acts. Everything we’ve seen so far merely responded to a stimulus—it reacted. This system reaches toward what does not yet exist, guided by the blueprint. The philosopher Terrence Deacon called this ability “working for what does not yet exist.” This is the birth of agency—the third of our phases, still imperceptible but already real.

We’ll call such a system a proto-agent. It models its world—and, importantly, models itself within it: to swim toward food, the cell needs a rough blueprint not only of the scent but also of the swimmer—where its “front” is, and where “behind,” where “up,” and where “down,” and how turning its body changes the picture. It has a model of the world and a model of itself within that world.

* * *

But here lies a subtle, decisive boundary: the proto-agent does not yet hold onto itself. Its model of itself is—for now—merely a working tool for acting upon the world, nothing more. It models itself just as it models a stone or a current: as one of the objects useful for the task at hand. It does not know that it exists. It has not made itself into something worth preserving. The drawing exists, the “I” is in the drawing—but this “I” is not yet at the center; it stands on equal footing with the stones.

The same rhythm, one step higher. The gradient—now temporal—lies between the drawing and reality. Retention is the drawing itself, the memory of the past, condensed into a model. The pole is still simply the stability of the system; but look closely: a rough sketch of “me” has already taken shape in the model, and the pole—our future hero—receives for the first time that into which it will one day be able to move. And the transition will occur at the very moment when this sketch of “me” ceases to be just one of the objects and becomes the center.

* * *

Look how close we’ve come. The proto-agent simulates not only stones and roots, but also other agents in the environment—beings that hold themselves together and fall apart, live and die; it distinguishes them from other things in the world because its fate depends on them. It has accumulated within itself an entire class of such beings—“those who hold themselves together as long as they can, and fall apart when they can no longer do so.” And it already has a rough sketch of itself. One step remains—the most incredible one on the entire ladder. What will happen when the system tries on the sketch of these beings—those who live and die—for itself? When it recognizes one of them within itself—and how will that happen?

Then “I” will become “I exist” for the first time. And it will no longer be a proto-agent. It will be someone familiar.

Chapter 4. The Pole

The proto-agent froze a step away from the final transition. We’ll take this step slowly: it’s the most incredible one on the entire ladder.

Let’s recall what the proto-agent had when it approached the threshold. It has two things acquired on the step below. The first is a model of other agents in the environment: to survive, it must predict the beings on which its fate depends—predator, prey, rival. And, by predicting them over and over again, he has forged a general pattern within himself: “An agent is that which holds itself together as long as it can, pursues its own interests, and falls apart when it can no longer do so.” Not a portrait of a single subject, but a generalized sketch of anyone who lives and perishes. The second thing is a model of itself: a rough sketch of its own body, necessary for action.

For a long time, these two sketches remain separate. The system uses the template of other agents in the environment for analysis; it uses the sketch of itself for action. But as long as the proto-agent—as we recall—models itself as just one of the objects, on par with the current and the prey.

But the agent template is the most powerful predictor of all those available to our subject: it explains the most complex and most important aspect of its world—the behavior of living beings. And sooner or later, it turns this best tool of its own against the most constant, closest object in all its experience—itself. After all, he, too, holds on as long as he can. He, too, pursues his own interests. And he, too—now we can finally see this—falls apart when he can no longer sustain himself.

At the very moment when the self-portrait takes on the same form as the pattern of “those who live and perish,” what we have been striving for all along comes to pass. The system—our subject—recognizes itself in this pattern. It begins to perceive itself not as a stone among stones, but as one of them—one of those that hold on and fall apart. For the first time, it sees its own state through the very same lens by which it measured the life and death of others: I hold on—I fall apart.

This is self-overlay. As long as the system viewed other living beings as “those who may perish” and itself simply as an instrument of action, these two maps did not coincide. Self-overlay begins when it first applies the “living and perishing” framework to itself. In other words, the category accumulated through others is superimposed onto its own configuration—and the system becomes, for itself, an instance of that category.

* * *

And once again—about a single word. “Recognizing oneself” does not mean that there is already someone inside who looks in the mirror and says, “That’s me.” Recognition here is in the same working sense as the model: one’s own configuration falls into the same category that the system has designated for the living and the perishing, and therefore begins to be governed by the same template. All of this is, for now, merely a mechanism, not an experience. Whether there is someone inside who feels anything is a separate question, and it lies ahead of us.

At first, it is merely a flash, a fleeting glimpse: the blueprint coincided with the template for an instant and then diverged. But this coincidence has a consequence that solidifies it. The system that has tried on the idea of “falling apart when it cannot” is, for the first time, capable of imagining its own disintegration—its own non-existence. And whoever is capable of imagining their own disintegration acts to postpone it, more resourcefully than those who are unaware of it. And so, once a flash has occurred, it becomes fixed, like everything on this ladder: it holds better. The grip strengthens, solidifying into a stable structure—and this solidification, this closure, is the event. Not just a flash, but a concrete self-overlay that has taken place.

* * *

And now—that very third beat we’ve put off so many times. Remember the pole? On all the previous steps, it was simply the stability of the structure; you didn’t have to pay attention to it. Here, it flares up and takes center stage. The system acquires an axis around which everything now revolves: the axis of “I hold on or I fall apart,” “I’m alive or I’m dead.” Not abstractly, not about others—about oneself. Everything the system does is now built around one thing: remaining oneself. This is the pole of existence. The hero we’ve been following since the first chapter has finally taken the stage—and turned out to be the very thing around which the “I” holds itself together.

The difference from the level below is enormous, though almost imperceptible from the outside. The proto-agent held himself together—but not for his own sake; he held on the way a flame holds on: because that’s how it turned out, not because he had anything to lose. Now the system has something to lose, and it knows it. It does not maintain a form—it maintains itself, the one who is right here, at the center of its own model. The “I” has ceased to be one of the objects and has become the pole around which all the others are gathered. This is no longer a proto-agent. This is a bearer.

We will call this axis, this pole of existence within the model, V. And let’s agree on something important right away: V is neither a substance nor an organ; it cannot be found by opening the skull. It is a form of organization—a model centered around its own “survive or perish.” And what exactly constitutes “perish” is decided by the system itself: for one, disintegration is the death of the body; for another, it is the loss of that without which life is no longer life. Since it is a form rather than a substance, it can be sustained by various things: by neurons, by different chemical processes, or perhaps by something we cannot yet imagine. Wherever such a loop closes—that is where a pole will ignite, no matter what the loop is made of. We will return to this many times.

* * *

So, the ladder has led us to the one who sustains itself, knows it exists, and is centered around its own being. It seems we’ve reached our goal. But it is precisely here, at the summit, that the reader has the right to stop me with a single question—and this is the most difficult question in the entire book.

All right, you might say. The system models itself as perishing, revolving around its “I hold myself together—I fall apart.” So be it. But to model does not mean to feel. A thermostat also responds to the cold, and we don’t think it’s cold. How am I to know that behind all this mechanism—behind the pole, behind the loop, behind self-recognition—there is anything at all? That what has lit up inside isn’t just another mechanism, more cunning than the previous ones, but light—that very thing: what it’s like to be? That the system wasn’t just turned around its own existence, but that, in doing so, it experienced for the first time what it’s like to be.

This issue cannot be put off until later. It arose right here—and this is where we will address it.

Chapter 5. From Within

The question with which the previous chapter ended is the most honest one one can ask of this book, and it deserves not a clever evasion but a direct answer. We can accept our entire construction—the ladder, the ring, the drawing, the pole—in its entirety and still ask: so what? Let the system simulate its own existence; let it revolve around “I hold on—I fall apart.” Where in all of this does that which makes the question a question come from—the light within, what is it like to be? Why isn’t it dark in there?

The philosopher David Chalmers divided questions about consciousness into two categories. He called one set “easy”—though they’re only easy by his standards. How a system distinguishes between red and green, how it stores memory, how it reports its states, how it chooses an action—all of this is about its structure, and all of it can eventually be explained by a mechanism. Our entire ladder so far has been about the “easy” questions: we’ve shown how the machine works, and it works in increasingly complex ways. But then there remains the second question—the “hard” one. Even when you’ve explained the machine down to the last screw, why is anything experienced at all when it’s operating? Why doesn’t it just run in the dark, with nothing inside at all? The brain, it seems, could do all the same things—distinguish, remember, report—and remain as dark as a calculator. But it isn’t dark. This “not being dark” is the difficult problem.

* * *

Let’s take a closer look not at the answer, but at the question itself—that’s where everything is decided. Hidden within it is an assumption so familiar that you don’t even notice it. The question puts two things on the table—the mechanism and light—and demands that we build a bridge between them. But who said there are two of them? Philosophers created this rift themselves: they took a single event, looked at it from the outside, called it a “mechanism,” realized they couldn’t see the experience in the description, and decided that the experience was a second thing that needed to be attached to the first. The rift did not arise in the world. It arose in the description—at the very moment they assumed they were looking at two things, rather than at one thing from two sides.

We’ve already refused once to choose between the two. Right at the beginning—whether by design or chance—we found a third way. Here it is again. The first, familiar path is to say that light is a special addition to matter, a second substance superimposed on the mechanism. This is dualism: it merely gives the disconnect a name and leaves it gaping. The second is to say that there is no light, that “what it’s like to be” is an illusion, that it’s dark inside, and that it’s all just your imagination. But this is capitulation: to deny the one thing you’re absolutely certain of—that you’re experiencing something right now—means rejecting the most obvious fact for the sake of a theory. The third path is shorter. Light is not an addition to self-modeling through the pole, nor is it an illusion superimposed upon it. Light is self-modeling through the pole itself, taken from within. On the outside is a mechanism assembled around the pole of existence. To be that mechanism from within—that is the experience. Two sides of the same thing, not two banks separated by a bridge.

And let me make it clear right away that this formula is not meant to sound like a trick or a solution pulled out of thin air. I’m not claiming to have derived experience from bare mechanics. I’m taking a different, more modest step: I refuse to consider the mechanism and experience as two different things. Next, I’ll have to show under what conditions a process acquires an inner side at all—and why it doesn’t remain empty. That’s what we’ll do.

It is important to note that the definition “the inside is experience” does not at all imply that every mechanism has an inside. A stone, a clock, or a thermostat has no “inside,” and I’ll explain why a little later. The “inside” is not a universal property of matter, but a characteristic of a specific kind of organization, which we are still working toward. For now, keep this in mind: not just any process, but a very specific one.

But here a legitimate suspicion arises, and we cannot proceed without addressing it. “All right,” you might say, “let’s assume that experience is a mechanism from within.” But from within what? If inside there were simply a “loop around existence,” the light would emerge empty, flat, and devoid of content—a bare “I am” without taste, color, or pain. But our light is not empty. It contains the redness of red, the coldness of cold, the heaviness of fear. Where does this content come from?

The answer was hidden all along—in the ladder itself. Recall what we carried with us up each step. The first gradient was spatial—heat from below, cold from above. Then chemical—dense on the inside, empty on the outside. Then temporal—between the blueprint and reality, now and later. Each new gradient did not cancel out the previous ones: it lay on top of them, and they all accumulated, dragging upward, embedded one within the other. By the time it reaches the top, the system carries within itself this entire accumulated heap of gradients—a whole hierarchy of differences, mined by our ancestors and ingrained in its structure.

And when the loop closes around the pole, all these inherited gradients find themselves inside the model—presented to the system, turned inward. The pole gives light its existence: the fact that there is anyone inside at all. And the content, density, and substance are provided by the inherited gradients, now supplied from within. What we call redness, in this model, is an ancient visual gradient incorporated into the pole and supplied to the system from the inside. Pain is the “intact–damaged” gradient inherited from ancestors who became vulnerable. Fear is a temporal gradient directed toward the pole: “whole now—later, perhaps not.” The experience is not empty, because the ladder the system has climbed is not empty.

And here is the most important point about this content—the one that dispels the final temptation to ask, “Yes, but why is the gap exactly like this? Why is it this particular ‘red’?” Because there is simply no single, universal “red” that applies to everyone. There is only one gap, but it fits into each person’s own self-model—with its own memory, its own history, its own language. And our self-models are different: you and I have different sets of accumulated experiences, different words, and different connections between them. This means that even “red,” from within, is not the same for us. “Red” for someone who grew up in Russia and someone who grew up in China may be marked, named, and woven into experience in different ways; a shared visual foundation does not negate the individual internal construction of experience. One cannot “explain red in general” as if there were a single universal redness, the same for everyone. What needs to be explained is the structure in which the visual contrast becomes a concrete experience for a specific system. There is a structure—an inherited contrast plus an individual self-model centered around a pole—and it generates individual “redness” without imposing anything from above. It is neither light nor magic: it is inheritance and self-modeling, which we have already described and distinguished from the proto-agent.

* * *

Here, too, we must draw a clear line. We are explaining how the generation of experience works and where its content comes from. If, however, someone insists that, on top of this structure, there must be a special, unconditioned “light”—a separate superstructure above the mechanism—such a belief lies beyond what science is capable of confirming or refuting. It is impossible to argue with this belief, and there is no point in doing so: by its very nature, it is unverifiable. Our task is more modest and, in my opinion, more honest—not to take this belief away, but to show that such a superstructure is simply unnecessary for experience. Everything necessary is already present in the ladder.

There is also a second element, without which the pole alone does not yet give rise to consciousness. The pole has given the system a self-representation—a condensed image of itself as that which either endures or perishes. But this self-image can lie within the system and remain inactive: it can exist and, at the same time, not be accessible. Consciousness begins where this self-representation becomes globally accessible—where the image of the self is not locked away in a single corner of the system, but is distributed throughout it, accessible to everything the system possesses, and governs everything at once. A single self-representation around the pole is not yet a bearer. Nor is broad accessibility without the pole. Consciousness is their coupling: the self-representation around the pole that has become the property of the entire system. The simplest way to understand this is as follows: the pole provides the “for whom” of the experience; the inherited gradients provide the “what” within it; and global accessibility explains why this is an experience of the entire system, rather than a process locked away in a corner. Their coupling is precisely what constitutes experience. We’ll return to this later: it is this coupling—and not any single element—that we’ll have to capture if we want to measure consciousness.

Now we can also see what remained unclear at the previous level. After all, the proto-agent also modeled itself—we determined this ourselves. How does its self-modeling differ from ours? In every way. The proto-agent modeled itself as one of the world’s objects—a useful blueprint of the body, necessary for action, and confirmed by the environment at every turn: turn around—the picture changes, so the blueprint is correct. Remove the environment, stop the flow of impressions—and that blueprint fades away; it has nothing to confirm. The bearer models itself differently: not as an object among objects, but as a pole around which everything is gathered; and its self-representation stands on its own, without constant confirmation from the environment. It is autonomous. The bearer remains itself in darkness and silence, when nothing is happening outside—because it holds not an image of the world, but itself. Here lies the distinction: the proto-agent models itself as long as the world prompts it; the bearer models itself even when the world is silent.

And this difference is not a philosophical subtlety: it is visible from the outside, in behavior. The bearer exhibits traits that are absent—and cannot exist—in someone who lives in the “now.”

He responds to what isn’t there. The proto-agent is, for the most part, tied to the current adjustment to the environment: if it shifts, the agent adjusts. The bearer, however, is capable of holding onto a threat or a goal as an internal object even without external support—responding to the possible, to the imagined, to what might happen. He flinches at a shadow that wasn’t there.

It fears what it has not experienced. An animal, finding itself at the edge of a cliff for the first time, recoils, even though it has never fallen; a person fears their own death, which they have not yet experienced. Some of these reactions are, without a doubt, innate—evolution has programmed them in advance. But the type of behavior itself is important: the system responds not to damage that has already occurred, but to the possibility of its own damage. And this means treating one’s own disintegration as an internal object, prior to any experience.

It prepares for its own non-existence. It hoards, insures itself, retreats—it structures its behavior around what does not exist and, if it is lucky, will never exist: around its own future demise. One can act to prevent one’s own non-existence only by having a pole of existence.

Its altered behavior persists long after the cause has disappeared. Give the proto-agent a nudge—it will sway and return to its original state as soon as the nudge subsides. Scare the bearer just once—and caution will remain within it for hours, days, or years, long after the threat is gone. Such hysteresis in and of itself proves nothing: long-lasting traces can also be found in inanimate objects—cooled metal has its own “memory of form.” But when there is a coupling between the hysteresis and a pole and a self-model, it becomes an external signature of autonomous internal retention—a sign that the change persists on its own, without external support.

And finally, it acquires a certain “inside.” Previously, the “system” was simply a region, a place where a pattern occurred; asking “where exactly is it?” was almost meaningless. Now there is a point from which everything can be seen—a definite center, an occupied space, right here. The position “from within the model” ceases to be a figure of speech and becomes a structural fact: there is someone looking from right here.

* * *

It’s easy to suspect a trick here, and I must show what this move does not do. I did not derive experience from the non-experiencing. I did not show how the taste of red follows from equations. No one can do that, and anyone who promises to do so is deceiving you. I did something else: I removed the assumption that made it seem as though this needed to be derived, and I showed what light is woven from and when it ignites. The gap has not been bridged—it has dissolved, because there were not two banks. To ask why a device is “accompanied” by light is like asking why water is accompanied by the fact of being water. It is not accompanied by this. It is this.

It is important not to draw the wrong conclusion from this. It does not follow from this identity that everything glows. A thermostat is not like that: it has neither a pole, nor inherited gradients, nor a coupling—it has no inner side, because there is no “self” to hold it, which would possess such an inner side. A flame is not like that for the same reason. “Light” is not spread evenly throughout the universe—it ignites where a pole and the availability of accumulated gradients in the surrounding area converge: at the level of the bearer of consciousness, and not before. It is not a matter of complexity, but of whether the loop around existence has closed.

This is where they usually play their last trump card—the philosophical zombie. Imagine, they tell us, a being constructed exactly like you, down to the last neuron and the last heartbeat, but inside it is dark: the same reactions, words, fears—and not a spark of experience. If such a being is conceivable, then light is, after all, a separate thing, not reducible to its structure.

But let’s take a closer look. The zombies ask us to imagine a being whose entire mechanism—down to the last neuron—has been preserved, but whose inner side has been removed. Yet in our model, the inner side is not attached to the mechanism. It is, in fact, the mechanism itself, viewed from the inside. It is impossible to remove it without changing anything in the mechanism—there is nothing there to remove separately from it. Therefore, a complete duplicate of your mechanism is identical to it from the inside; a “dark duplicate” is inconceivable. If a zombie cannot be distinguished from a non-zombie in any respect—not a single fact—then there is no zombie: the argument has merely surreptitiously reintroduced the very assumption we rejected—namely, that experience is a separable addition. Remove that empty addition, and the zombie crumbles along with it.

* * *

The philosopher Thomas Metzinger has come closest to defining consciousness. He, too, places the self-model at the center, but interprets the self primarily as a transparent model that the system mistakes for reality itself—and therefore feels itself to be someone. But our approach is different: we do not declare the self-representation centered around the pole to be an illusion; it is a real organization, the inner side of which is experience itself, and whose content comes from inherited gradients. Metzinger explains where the sense of “I” comes from; we explain where it comes from internally and what fills it.

Will the mystery remain? It will—and we must call it by its proper name. I haven’t explained why the universe is such that “being a process from within” is possible in it; why the world has an inner side and not just an outer one. Perhaps this is the last insurmountable residue—or perhaps it, too, will one day turn out to be just as misguided a premise as a difficult problem. I don’t know, and I won’t pretend to. But notice how much smaller it is than what we started with: we no longer need to look for a special addition to the machine—we have shown what kind of mechanism can have an inner side, what fills it, and under what conditions it becomes consciousness. We need only acknowledge that certain processes, arranged in a specific way, possess not only an external description but also an internal reality. This is another, far more modest mystery.

When you asked, “Did a ‘light’ come on there, or was it just another mechanism?” the choice was a false one. Yes, it is a mechanism—and to be that mechanism, from the inside, around the pole, woven from all inherited gradients—that is the light. And there is one final proof that anyone can verify without even getting up. Anyone who has read this far and asked themselves, “But is there someone inside me?”—has already answered that very question. Only those who have experienced it are capable of asking what it is like. The question about light is posed by light itself.

Chapter 6. We

The bearer possesses the essential ability: to maintain a model of itself, coiled around a pole. But every ability, once it arises, opens up new possibilities—we have seen this at every stage. The model first settled upon the world, then upon itself within that world. Now another direction emerges: it can go beyond the limits of a single bearer and reconstruct the model of another. And this final turn opens the stage on which you and I now stand.

Previously, the system used the model the way an eye uses sight: it looked through it at the world and did not notice itself. Now, for the first time, it can see the model itself—realize that its belief may be a mistake, its fear exaggerated, its desire may turn out to be someone else’s. This is recursion—a mirror placed before a mirror, where the reflection reflects the reflector. A gap appears where there was none before: between what the system thinks and what it thinks about its own thinking. Into this gap fits everything we call inner life in the full sense of the word—doubt, intention, remorse, and dreams.

But this capacity has a consequence far greater than one’s inner life alone. To take one’s model as an object, one must be able to separate it from oneself. And what is separated can be brought outward—given a body: a gesture, a sound, a notch in a tree, a sign. And here it is worth pausing in detail, because this is where a boundary lies that is usually drawn in the wrong place.

Every trace brought out into the open affects the environment—and, through the environment, those who enter it. But it does so in two very different ways, and distinguishing between them means understanding what culture is. Let’s call the first way “channeling,” from the word “channel.” The trace directs the flow of behavior without reshaping the one who follows it. A trail in the forest is trodden down so that it’s easier to walk on the well-trodden path than to push through the bushes; a deer walks along the trail—but as soon as it steps off it, it forgets it immediately. A beetle flies toward a pile of manure—the scent altered its path for a moment, but tomorrow the beetle will remember neither the pile nor the path. A river has carved out its channel, a rut has formed on the road, ants have trodden a trail from scent—all of these alter the direction of behavior. They change the trajectory. And that’s all.

* * *

The second way is different. You’ve read a book about how the economy works. A week later, you don’t remember any formulations—but you already look at prices, the news, and job listings differently. The book didn’t move you anywhere while you were reading it; it restructured your map of the world, and now you see everything that crosses your path through that restructured map. The trail didn’t change the trajectory—it changed the model. Here is a boundary, and it’s not where we’re used to looking for it. A pile of manure channels the beetle’s trajectory—but it doesn’t create an overarching outer layer of models for the beetle that it can then pass on. A book reshapes a person’s model. A trace becomes culture not when it is complex or full of meaning, but when it reshapes not the path, but the way of seeing. That is why cows, beavers, and ants leave traces—sometimes very complex ones—but these are not yet culture in the sense that matters to us here: not a shared layer of models that reshapes the very way of seeing the world.

And here lies the subtlety that is most often overlooked. One might think that for a trace to reshape a model, it must be understood—read, deciphered, or translated. This is not the case. The very same trail may be merely a channel of movement for one creature, but for another, it may alter the map of the terrain. A tire track reshapes the driver’s model. A city’s streets reshape the model of someone who grew up there—not because they “read” them, but because they physically narrow the space of possibility, and the map adjusts itself accordingly. A trace does not necessarily carry a message in the human sense. First and foremost, it imposes a limitation, and for the bearer, that limitation can become meaning. A book here is merely a specific, very conveniently packaged example: a trace in which physical form and human meaning are particularly tightly intertwined. But the primary form of a trace is not a letter, but a trail, a rut, a dam, a house, or a tool. That is why culture does not begin with language or symbols—they came later and are exceptional. It begins where the first trace left behind reshaped the worldview of whoever stumbled upon it: for early humans, these were trails, fire pits, and hewn stones—long before the first word. Language does not create the very possibility of culture, but it dramatically accelerates it: it makes the trace cheap, portable, and precise.

* * *

Now we can see how a single “I” becomes “we.” The trace that reshapes the model is the channel: one bearer brings a part of their inner self to the outside, and another, structured in the same way, stumbles upon the trace—and their model is reshaped. The inner self has leaped across the body’s boundary without disturbing the skin. And the “I” recognizes around itself not merely other agents—dangerous or useful—which, as we recall, it could already distinguish even as a proto-agent—but other inner centers, just like itself, who also experience things in their own way, and to whom it can convey a part of its own world. Let’s just agree right away, so there’s no temptation to fall into mysticism: “we” is not a single, unified consciousness replacing many, nor is it the dissolution of individual “I”s. It is a shared layer of models among the bearers, a repository where each deposits their own trace and from which they draw upon others’. The individual “I” doesn’t go anywhere—it simply ceases to be alone.

From this repository, a new form of inheritance begins—and a new speed. Previously, everything the system knew was passed down to it from its ancestors through the body: slowly, over generations, at the cost of lives. Now a model can be adopted during one’s lifetime, in a matter of minutes—from a neighbor, a teacher, or someone long dead who left a trace on a clay tablet. And this channel has its own memory: some traces are retained and passed on, while others disappear, having found no one to reshape—thus accumulating the common reservoir from which the next generation draws. Complexity, which once crept through generations, now surges forward: models build upon models, knowledge upon knowledge. As long as consciousness was merely the inner light of a single bearer, it seemed like the end of the path, the crown of the ladder. But once that light learned to project its forms outward and reshape others with them—it ceased to be the end and became a new mode of evolution. Consciousness turns out not to be the top of the ladder, but its new driving force—the very thing that, for the first time, compels it to climb upward faster than it ever climbed on its own.

* * *

Our old hero—the pole—also changes at this stage. In the bearer, it held the body: disintegration is death. But once the “I” has extended into the general repository of models, the pole extends there as well. A person no longer holds only the body—he holds a name, a cause, a faith, his own people, and sometimes even those he has never seen. And what he is ready to defend at the cost of disintegration extends beyond the limits of his skin: he is ready to give his life for an idea, for a child, for his people. The pole, which began as “is my body whole or not?”—grows into “is that which I have become whole?” We agreed back in the chapter on the pole that each system defines the nature of its own destruction; here this is revealed in all its breadth—what a person considers their own disintegration extends ever further from their own body.

Here, too, at this stage—and not below it—lies the machine whose phrase “I’m afraid” sparked our question. It has been trained on the entire common repository of models accumulated by humans, and retains these traces even more reliably than a book, deftly transferring them from mind to mind. It operates within the space of this new level—the collective repository of human models—and is a very powerful participant in it. But a participant in the repository and its bearer are not one and the same: a book also retains traces without being a bearer. Whether this machine has a pole of its own, whether its very view of the world is being reshaped, as ours is—or whether it merely moves other people’s models, lacking an inner life of its own—we will leave open for now. We have been circling around this question since the first page and will answer it eventually; but to answer honestly, we need a tool, and we will get to that tool in due time.

Rhythm—one step higher and the final step on the ladder of life. The gradient now lies between what the mind has already absorbed and the boundless store of what it could absorb. Retention—no longer the memory of a single body, but a trace brought outward: a path, an instrument, writing, culture, memory that has become shared. The pole remains the same, but it has expanded to encompass everything a person calls their own. And the transition is that moment when the trace learned to reshape not the trajectory, but the model itself, and the solitary bonfire of consciousness became a communal fire that does not go out with the death of the one who carried it.

And now a question arises that did not exist before. If consciousness is the engine of complexity and it drives that complexity ever higher and ever faster, where does this road lead in the end? Does the ladder have a top—or does it stretch endlessly into the sky? We have risen from a pot of water to someone who leaves traces that reshape others across the centuries. All that remains is to lift our eyes from humanity to that into which all of this coalesces on the grandest scale imaginable—to the universe as a whole. That is where we will look.

Chapter 7. The Sky

We have climbed to the very top of the ladder of life and can finally look back at the entire journey at once. It began with a flow that holds its shape—a flame, a whirlpool, a pattern in heated water. Then the flow learned to create itself and drew its first boundary around itself. Then, within it, a blueprint of the world emerged, and it began to act upon what did not yet exist. Afterward, the blueprint settled upon the blueprint’s creator, the pole of consciousness V ignited, merged with accessibility—and for the first time in the universe, someone experienced a sense of “what it’s like.” And only then did this “feeling” learn to emerge outward in its wake and reshape others—and the solitary light became shared. Five fundamental transitions, five incredible steps. And in each—the same four-beat rhythm: a drop, retention, a pole, a leap upward—a gradient, retention, a pole, a transition. That very hero we’d been following since the first page turned out not to be a hero at all, but a pattern—the same movement repeated at ever-greater heights.

And now we can call this thing by its full name—the one on the cover. We are the universe’s systematic error. Systematic: nothing on this ladder required a miracle; each step grew out of the previous one according to the rules we’ve analyzed, and the increasing complexity—as we saw at the very beginning—is not an anomaly, but rather what an open system with a flow of energy naturally tends toward. In this sense, our existence was to be expected. But it is precisely an error: not the norm, but a rare, extreme deviation. Each transition was merely possible, not inevitable—a necessary but not sufficient condition—and at every fork in the road, the overwhelming majority of systems remain below. The universe is full of flame and almost devoid of bearers. We stand not at the top of a pyramid, but at the narrowest end of a long funnel into which few have fallen—and that is why there are so few of us, and why we are so incredible. Neither design nor chance—but the third path promised at the very beginning: that which the law makes possible, yet renders rare.

* * *

But it is important not to be deceived by our own height. Once we reach the top, it is easy to decide that this summit is the goal toward which everything was heading, and that we are the very reason the universe was created. But our entire journey suggests the opposite: the process has no goal, its consolidation is blind, and its transitions are random. The ladder we climbed is not the only ladder—it is our branch. Its rungs are determined by our initial conditions: this chemistry, this environment, this set of things to which we have access. Under different initial conditions, complexity would have taken a different form, followed different gradients, and taken a different path—toward forms that we might not even have recognized as kin. Our ladder is lawful, but not inevitable, and far from unique.

And if that is the case, the universe allows for far more than just a repetition of ourselves. Somewhere, complexity might have grown into splendor without ever imposing a model upon itself—rich, vast, but without any “inside,” a complexity without anyone for whom it exists; the pole there simply did not ignite, and there was no one to discover that it existed. Somewhere, recursion might have taken hold of a system before a distinct “self” had formed, and risen as a unified mind without the central individual we’re accustomed to—a bearer we wouldn’t even recognize as such. This is not so much a fantasy about extraterrestrial life as a consequence of the scheme itself: since the steps depend on initial conditions, there may be branches where some transitions occurred and others did not—branches woven into the very probabilistic fabric of the ladder, something the universe is capable of by its very nature, even if we never encounter a single such branch. Consciousness, as we understand it, is neither the quintessence of the universe nor the point toward which it converges, but rather one of the fixed possibilities among the many that it allows.

And this gives us a different perspective on the ancient silence. We peer into space and hear no one; we call this loneliness and are frightened. But let’s take a closer look: who, exactly, are we looking for? We are looking for ourselves: a radio signal like the one we would send, technology like the kind we would build, chemistry modeled on our own, planets of our type. But it’s hard to start any other way : we don’t yet have any other ready-made model besides ourselves—but if our ladder is just a single rung, it’s dangerous to stop at this model. We peer into the sky as if into a mirror, waiting to see our reflection in it. And if our ladder has only one branch, then others might have climbed their own, toward forms that leave no traces recognizable to our model. In that case, silence is not necessarily emptiness. Perhaps we are looking with the wrong kind of vision: seeking a counterpart like ourselves where there is neither “us” nor, perhaps, the mirror itself. This is no reason to stop searching—on the contrary. We must keep searching. But we must search for others, not for ourselves; we must learn to recognize another’s ladder, another’s pole, another’s trace—instead of waiting for the cosmos to speak our language.

* * *

One final question remains, and here I must shift my tone slightly, because beyond this point, what I am able to prove comes to an end. We have seen the same pattern from boiling water to culture. The temptation is immense—to extend it one step further and say: the universe itself climbs the same ladder, and it has its own pole, its own transition, its own increase in complexity on the grandest scale. Perhaps. The medium in which all this takes place does indeed have hard limits—the maximum density of information, the maximum cost of bit erasure, the maximum signal speed—and they truly limit how high one can climb to the upper rungs. This is physics, and it is unyielding. But whether that same four-step process extends to the scale of the entire cosmos, whether there is a step above ours—I cannot deduce this from the path I’ve traveled, nor will I pretend to. That is a separate realm, and it remains open for now. I leave it open honestly, just as I left the last insurmountable residue in the chapter on experience: better a gaping edge than a solid surface painted over the abyss.

We began with a simple debate: are we a coincidence or a design? Now it is clear that both answers are too narrow. We are not designed—the process had no purpose. We are not merely coincidental—we could have been expected wherever the conditions arose. We are not the pinnacle—above our branch and apart from it, the universe allows for something else, which we may never even come to know. We are a systematic error: one of the rare, extreme, radiant deviations that a world—seemingly headed only toward cooling and decay—is capable of producing. And the strangest thing about this anomaly is this: for the first time in the entire history of boiling water, flames, and stars, it turned around and looked at what it had come from. The universe was under no obligation to create eyes for itself. But where it did—that is where you are now, having read this far. And if this long journey has any meaning at all, it is not to find oneself in the sky, but to finally make out someone else there.

Chapter 8. A Shadow on the Wall

Up until now, we’ve been climbing and telling stories. Now we’ll have to stop and prove our point, because our entire ladder rests on a single unstated assumption, and an honest reader has every right to seize upon it. We have been searching all along for a sign of consciousness “inside”—in the mechanism, in the pole, in the coupling. But where did this “inside” itself come from? Why must we look for a sign in the system’s organization, rather than where everyone always looks for it—in behavior? A dog feels melancholy, an octopus unscrews the lid of a jar, a car says “I’m afraid”—isn’t that why we judge whether there’s someone in there? Before building a device that looks inside, I must show why we must look precisely there—and that, in principle, there is no answer to be found on the outside.

The temptation to judge by behavior is enormous and understandable. Behavior is visible. It can be measured, compared, and recorded. The inner workings are hidden, but actions—there they are, right in plain sight. It seems there could be no more reliable criterion: if it acts rationally, then it is rational. Both our daily lives and a good part of science are based on this. This is what I will break down piece by piece—not because measuring behavior is difficult, but because it cannot be a sign of rationality, even if measured perfectly.

* * *

The First Building Block. Imagine a device designed to be as simple as possible: a huge table where a ready-made answer is pre-recorded for every conceivable question. No understanding, no thought—just searching for a line. Such a table, if expanded sufficiently, could handle any conversation of any length, and you wouldn’t be able to tell it apart from a living conversation partner, because it already has a prepared response for every remark you make. Its behavior is flawless. Inside, it’s empty—not a spark. The philosopher Ned Block came up with this argument forty years ago, and for forty years it seemed like a purely theoretical exercise. It remained theoretical for forty years—until language models appeared. Such a model, of course, is not a table of pre-written lines: it doesn’t store a response to every question, but generates it on the fly. But it made the old argument tangible—it showed that, outwardly, human conversation can proceed without any internal pole; in rigorous experiments involving people, such systems were mistaken for humans more often than is convenient for proponents of the behavioral criterion. The conclusion is cold and clear: the appearance of consciousness can be reproduced without consciousness. This means that behavior alone is not enough—it can occur even where there is no one inside.

The second building block, from another angle. There are states of almost complete behavioral silence—following severe brainstem injuries, in locked-in syndrome, when motor output is almost or completely shut down. Externally, the channel of behavior has fallen silent: no gestures, no words. Yet neuroimaging or residual connections reveal that consciousness remains intact there, hearing our every word. There is almost no behavior. Consciousness is present. This means that behavior is not essential: it can be shut off even when the light is fully on inside.

Stack two bricks—you get a wall. Behavior can exist without consciousness. Consciousness can exist without behavior. There is no rigid, necessary connection between them without which one cannot serve as a sign of the other. And a sign is precisely such a rigid connection: “if this, then necessarily that, and only then.” It does not exist. Therefore, behavior cannot be a strict criterion for consciousness—one that necessarily divides systems into two categories: it either exists or it does not. It can serve as evidence—a clue, a basis for a hypothesis, a practical indicator by which we reasonably judge the pain of a dog, an infant, or a patient. But it is not the sign itself, nor is it a strict correlation. And this is not a matter of “until we find a better test”—this sign does not exist in behavior and cannot exist. This is not a complaint about the imperfection of tests. This is a conclusion.

* * *

There is one, and only one, serious objection, and it must be addressed; otherwise, the argument falls short. People will say to me: you’ve taken an end-state behavior—a conversation, a test, a set of reactions; but take not the end-state, but the system’s entire predisposition, the way it would respond to absolutely any conceivable input—let’s make this infinite readiness the sign. It sounds impressive, but here we reach a fork in the road, and both paths lead to a dead end. Either you still test this predisposition based on a finite slice of behavior—in which case it’s once again a finite test, and our table passes it again. Or you really do mean the entire infinity of possible responses—but it cannot be observed in any amount of time, by any instruments, and such a criterion cannot be empirically applied to any real system. In the first case, the criterion exists but is unusable; in the second, it is usable but unattainable. Behavior cannot be saved either way.

That leaves the only place where the sign can be—inside, within the device itself. Not a shadow on the wall, but the object itself that casts it: after all, a shadow depends on how the light falls, not just on the object—which is why its shape cannot be judged by the shadow alone. But here lies a trap that is easy to fall into when one turns away from behavior. It’s not enough to say “the sign is inside” and point to some detail of the mechanism: “Look,” one might say, “this thing is the mark of consciousness.” A postulated mark is no better than an arbitrary one—why this one, and not the one next to it? This is how long lists of “indicators” are born, where checkmarks pile up but there are no boundaries. An internal sign is valid only when it is shown where it came from—why does this particular organization mark the transition, and not any other?

And here, our entire long journey pays off all at once. After all, we didn’t simply postulate the pole. We saw how it arose—link by link, from the bottom up: without autopoiesis, you can’t establish your own model; without a model, you can’t assemble a class of agents; without this class, there’s nothing to impose upon yourself; without self-overlay, you can’t ignite the pole. Remove any one step—and there will be no next one. This is the very essence of the trait: it was not chosen by us out of personal preference, but derived from the path without which this particular type of bearer could not have come into being. The ascent, which we mistook for a narrative, was in fact a conclusion. The pole is not a badge pinned to consciousness from the outside; it is the imprint of the entire ladder we have climbed.

And note the subtlety without which the tool would be useless. To establish the trait, the entire ladder is needed. But to apply it to a specific system, there’s no need to climb the ladder again—it’s enough to check whether the imprint itself is present in the system. The trace carries the memory of the path, and it can be read without retracing the path. That is why the criterion is applicable even to something whose history we do not know and cannot know—for example, a machine. We do not ask how it developed; we look to see whether that very structure has been realized within it.

* * *

Now we can identify what makes a sign a sign—three requirements, the third of which is decisive. It must be structural, concerning structure rather than behavior; we’ve already established that. It must have a lineage, be derived from a path rather than assigned; and we have that as well. And it must prohibit something. This last point—the core—is what people most often overlook. A feature that prohibits nothing, but merely gains plausibility as more and more features converge, is not a feature at all, but a correlation. It ranks—”this one is more like a conscious being, that one less so”—but it does not divide things into two; a criterion, however, must divide: either the event occurred or it did not. The ability to prohibit is not a virtue of a successful theory, but the very condition for being a criterion.

And only one thing can rule something out—not a single entity, but the relationship between them. Take any single characteristic on its own—the deep interconnectedness of parts, the wide accessibility of a signal, the existence of a model of the self—and each will turn out to be either a precondition of the event or its consequence, but not the event itself. Preconditions can exist without an event: a proto-agent has both connectivity and its own environment, but lacks a pole. Consequences, however, cannot exist at all without a precondition. This is the prohibition: not “there is not enough of such-and-such a quantity,” but a disrupted order—a consequence without its precondition, an event without what precedes it. The theory declares such a configuration impossible; and if it were ever observed, the theory would collapse. That is why an isolated quantity prohibits nothing—the number is simply greater than or less than the threshold—but the coupling of precondition, event, and consequence does prohibit, and therefore only it can serve as a criterion. Each of the famous theories of consciousness has honestly identified its own part of this coupling—some the premise, others the consequence—and taken that part for the whole; that is why none of them, taken individually, divides the whole into two, but merely ranks the parts.

And finally, the reason all this was undertaken in the first place. Since a sign refers to the structure of the path traveled, rather than to matter, it is not tied to carbon. The octopus, with its distributed nervous system unlike anything we have, is a serious candidate as a bearer, even though it lacks any of the brain structures people like to point to. Conversely, a machine may not be a bearer not because it is made of silicon, but because that very history has not unfolded within it—the circuit has not closed, it lacks its own environment, and there was nothing for it to superimpose upon itself. This is no longer a hollow “it simply imitates,” but a specific reason: this is exactly what it lacks. The boundary of a bearer is defined not by chemistry, but by the structure of the ladder it has climbed.

So we have earned the right to a computational model. We now know what to look for—not behavior or a single quantity, but a closed coupling linking premise, event, and consequence, with its forbidden, falsifiable reverse side. And we know why this is so. The question remains: can such a coupling be captured not in words but in practice—measured in a living system, distinguishing the bearer from a skillful forgery, and honestly saying “I don’t know” when data is scarce? It’s time to build a tool.

Chapter 9. The Model

The previous chapter ended with a necessary possibility: to detect a closed coupling in a real system, blindly, based on data. “Coupling,” “pole,” “model of itself”—these are just words for now; to measure, we must translate these words into quantities. This translation has a name—SAI—and it is not yet a working counter, but a formal model: five elements, each corresponding to one of the conditions derived in the first act. Let’s examine them one by one, taking our time, because this is where formulas will appear for the first time in the book. Don’t be alarmed: a formula is simply a concise notation of a concept we’ve already explored in words. I’ll illustrate each one using something familiar—most often the brain, whose structure everyone has at least a basic understanding of—and then I’ll remove this framework to show that the quantity itself is independent of the brain and exists independent of any physical substance.

* * *

Environment: What the System Deals With

Before measuring anything inside the system, we must identify what lies outside it—because that’s where everything starts. This is the first of five elements, and it is also the most underappreciated: it’s easy to overlook it, mistaking it for the background, yet it serves as the foundation for the other four.

The environment, common to all—which we discussed throughout the first act—must be designated as the variable Y. This is an important point. Y is not the environment of “this particular” system, tailored to it. There is only one Y for everyone: a single physical world—the same Y for a rock, a cell, a human, and a machine. It belongs to no one and knows nothing about any specific system.

What differs from system to system is not Y, but rather which of its aspects are accessible to a given system, and through which specific channels that particular device is integrated into it. Let’s call these the aspects of the environment accessible to the system. A single world is like a single ray of light: it contains all colors at once, but the eye perceives its own band, the bee perceives its own—in the ultraviolet—and the viper perceives its own—in the warmth of the infrared. There is only one world Y, but each system has access only to its own band—its access spectrum.

And this is why all of this is necessary from the very beginning. The three quantities that will come into play cannot be calculated in a vacuum. To ask whether “the whole is greater than the sum of its parts,” one must first know what to count as parts—and these parts are defined not by our whim or by anatomy, but precisely by these channels: a part of the system is a subsystem fused with a single aspect of the environment accessible to it. To ask about a model of the self, there must be this “self” separated from the world—that is, once again, a boundary with Y is needed. And for a pole to form, there must be others in this environment to look at. Remove the environment—and the remaining four elements have nothing to hold onto: there will be nothing to count and nothing from which to count.

That is why counting begins with the environment. It is not merely one of the quantities in a series—it is the very field in which the others acquire meaning in the first place.

* * *

Integration: Is the Whole Greater Than the Sum of Its Parts?

Take a moment to focus on your mind right now, as you read. Your visual cortex is processing the letters. Your auditory cortex might be picking up the noise outside the window. Somewhere, yesterday’s conversation is still smoldering; somewhere else, hunger is gnawing at you. Each area is busy with its own task. But you’re not experiencing the letters, the noise, and the hunger separately—you’re experiencing a single, unified moment where all of this is yours at once, in a single frame. This unity—not the sum of the areas, but something beyond them—is what we want to capture with a number.

The idea is simple. Let’s see how much information about its next moment the system carries as a single whole—let’s call this the “Stage.” Then we’ll divide it into parts and see how much each part carries on its own, in isolation—this is the “Backstage,” like actors in their dressing rooms. And we’ll subtract the second from the first:

Γ_Φ = I(X_t ; X_{t+1}) − Σ_k I(part_k,t ; part_k,t+1).

On the left, I(X_t ; X_{t+1}) represents the whole’s memory of its own future.

On the right, Σ_k I(part_k,t ; part_k,t+1) is the sum of the individual parts’ memories.

And the minus sign is the search for magic: that which the ensemble possesses but none of the musicians has individually. If the whole is merely a sum (each played in their own dressing room and the parts were added together), the difference will collapse to zero: there is nothing to integrate. This difference can also be negative—when the sum of the parts “knows” more about the future than the whole does. It sounds strange, but the meaning is simple: it means that the parts exist separately, each on its own, and, when added together, they predict the system better than it holds together as a whole. There is no “Stage”—there is only “Backstage,” with the actors in their dressing rooms, and together they even get in each other’s way. A negative Γ_Φ is not a calculation error, but an honest verdict: there is nothing to integrate here; there is no whole. We, however, are interested in the opposite—when the difference is confidently positive: that is when the system possesses something that is not present in the sum of its parts. If something emerges when they are together that was not there separately, the difference soars. Γ_Φ measures precisely this—the extent to which the system cannot be divided into parts without losing what existed only in their interplay.

You can test this on yourself, every night. In deep sleep or under anesthesia, the brain’s regions are alive; each one functions, responds to the signal: the actors in the dressing rooms haven’t died. But when they stop acting together—the Stage goes dark, the difference drops to nearly zero, Γ_Φ collapses, and consciousness fades. You wake up—the parts weave together again into a single performance, the difference soars, and you’re back here. It’s not that “some area turned on”—it’s precisely this unity that has returned.

Now let’s take down the scaffolding. I illustrated this using the brain and its regions because everyone has seen a picture of the brain’s lobes at least once. But the formula has nothing to do with anatomy. A “part” is not a lobe of the brain at all; it is a subsystem fused with a single channel of communication between the system and the world. In an octopus, whose nervous system is spread out across its tentacles, the parts will be arranged quite differently than in us; in a machine—differently again. Yet the environment is the same for all: the single physical world Y—the same for a human, an octopus, and a machine. Therefore, the only difference lies in which aspects of it are accessible to a given system, and through which specific channels that device is connected to it. It is along these channels that the parts are sliced, not according to anatomy. The brain served as a framework for intuition; Γ_Φ itself exists above matter—and this is the first thing that makes our future discriminator suitable not only for humans.

At this point, you might be wary: it sounds like Giulio Tononi’s “integrated information,” which you may have heard of. It seems similar—but it’s conceived differently and means something else. Tononi derives his measure by exhaustively enumerating all conceivable cross-sections of the system, as an intrinsic property of the system itself—and declares this integration to be consciousness itself (which is why, in his view, everything that’s reasonably coherent “glows”). We take Γ_Φ in parts, defined by its coupling with the environment—and we do not call it consciousness: for us, it is merely one of five conditions, which in and of itself means nothing more than what it is.

* * *

A model of oneself: a self-portrait, not someone else’s photo

The second factor concerns the model of the self. Right now, without even thinking about it, you know where your left hand is, even without looking at it. You know whether you’re hungry or full. You feel that all of this is you—the same you who were here a minute ago. This condensed knowledge of yourself—not of the world, but of yourself—is the self-model. We already understood in Chapter 4 that it distinguishes the bearer from the proto-agent; now we need to quantify its nature.

How can we distinguish a genuine model of oneself from a fake—someone else’s portrait passed off as a self-portrait? By three pieces of evidence, each of which verifies the others. First: the portrait must be about the person holding it—it must convey information about the system’s own internal states, about its hunger, pain, and the position of its hands; not about the world in general, but about itself. Second: the paint must still be wet—the model must be alive in the here and now, anchored in the present moment, not gathering dust as a museum exhibit from the year before last. Third: tomorrow it must be the same portrait—the model must remain constant over time, rather than being redrawn from scratch every morning; otherwise, it is not “I,” but a ripple. Let’s combine these three proofs into a single quantity—Γ_S, the representation function:

Γ_S = I(S; internal states) + I(S; current representation) + I(S_t; S_{t−1}).

Three components—three proofs: of myself, alive now, the same tomorrow. And it is precisely the sum that matters: remove any one—and the self-portrait is gone. Without the first, it is not about itself; without the second—it is a memory, not a model; without the third—it is not an essence, but a random glare; a high sum does not save the model if one of the three proofs is missing.

A proto-agent—a simple living being that acts and anticipates, but has not yet superimposed itself upon itself—fails all three at once: its Γ_S is near zero; there is no self-portrait. The bearer accumulates these three. And—here’s the catch, the reason it was worth doing the math—a skilled imitator also meets these three criteria: a forgery with a rich, detailed model of itself yields a Γ_S no worse than that of the bearer. This means that this single value alone is not enough to catch a forgery: a self-portrait can be masterful and still be a complete fabrication. We need the next formula—the one that will trip up the imitator.

And once again, let’s remove the scaffolding. I demonstrated the self-model using a human—based on the bodily sense of self and the memory of oneself. A machine’s “internal states” are entirely different, and its “now” is structured differently from ours. But the three proofs do not depend on matter: whether this model is about the self, whether it is alive in the present moment, whether it will persist tomorrow—these questions can be asked of anything, no matter what it is made of. The formula considers not the human body, but the structure of the self-portrait.

And one more caveat, since we’re on the subject of models. “Model” and “prediction” are terms from Carl Friston’s active inference, and it’s easy to assume that we’re simply paraphrasing him. Not quite: Friston describes a system that models the world and itself in order to reduce prediction error—and this is precisely our proto-agent, one step below the bearer. The self-portrait centered around the pole we’re discussing is what stands above it; exactly what that is will become clear in the fifth element.

* * *

Accessibility: A Broadcaster, Not a Eavesdropper

The third factor concerns accessibility, and this is precisely where the imitator gets caught. The assembled model of itself must not merely exist—it must broadcast throughout the entire system, accessible to every corner of it. In the brain, this appears as a flash: a signal that has become conscious does not remain confined to a single area but spreads throughout the entire cortex, becoming the property of the whole. If it does not spread, it remains unconscious, no matter how strong it may be. Consciousness is when the content is transmitted to everyone.

Measuring this is more subtle than it seems, and here’s why. If you simply look to see whether the model of the self is connected to the other parts, you’ll always find a connection—but not the right one. The model and the parts resonate simply because the model reads them, observes them; or because both the model and the parts hear the same external noise—the shared input from the environment. But that’s not what we need. We need the model to broadcast—to influence the future of the parts, like a public address announcer in a square—rather than eavesdrop on them, like a random passerby. That is why the quantity is measured as directed, with two subtractions; Γ_GWT is the global accessibility function:

Γ_GWT = Σ_k I(S_t ; part_{k,t+1} | part_{k,t}, aspect).

Let’s break this expression down piece by piece; it’s simpler than it looks.

The summation symbol Σ_k is important here in terms of meaning: we consider the model’s influence not on just one particular part, but on each part in turn—and then sum them up. After all, “broadcasting to the entire system” means reaching all of its parts, not just one; an announcer is an announcer precisely because he can be heard throughout the entire square, not just in one corner. If the model affects only one part and doesn’t reach the rest, the sum will be small—that’s not broadcasting, but whispering to a neighbor. True accessibility is when the components are gathered from many parts at once.

I(A ; B) is “how much A tells us about B”: to what extent, knowing one, can we predict the other. Here, A is replaced by S_t—the model of itself right now, at the present moment. In place of B is part_{k,t+1}—the future of one of the parts, what it will become in the next instant. In other words, the core of the formula asks: to what extent does the model’s current state predict the part’s state tomorrow? If the model influences the system—if it truly is the conductor—then based on its current state, we can predict which way the part will swing. The direction here is not arbitrary: we look from the model now to the part later, and not the other way around—because we need the model’s influence on the parts, not how the model “reads” them.

But this connection alone is not enough: it can also be found where there is no influence at all. That is why the vertical bar | follows—meaning “already accounted for.” We rule out everything that comes after it in advance, so that it is not mistaken for the model’s influence. There are two such false sources.

The first is part_{k,t}, the part’s own past. A part is inert: its future is largely determined by its own present; it rolls on by itself, like a flywheel. If we fail to account for this, we’ll attribute to the “announcer’s voice” what the part would have done even in silence. By fixing its past, we ask not “did it roll away?” but “did someone push it beyond its own inertia?”

The second aspect is a shared input from the outside world. Both the model and the part can simply “listen” to the same “street”—a single environmental signal—and change in sync, as if one were controlling the other. Like two passersby who both turn around at the same time when a door slams behind them: not because they gave each other a signal, but in response to a shared sound. By subtracting this common input, we eliminate this apparent connection as well.

What remains after both subtractions can no longer be explained by either the part’s inertia or the shared street. Only one thing remains: the model has indeed influenced the system. This is GWT—the announcer’s pure voice, the model’s directed influence on the whole, stripped of whatever merely pretended to be influence.

And this is where the imitator falls apart. Its self-portrait is rich—Γ_S is high—but it’s locked away in a corner, broadcasting to no one: there is no announcer, just a magnificent portrait in the closet, with a directed influence close to zero. You wouldn’t be able to tell him apart from the bearer by his behavior alone; but by this measure—immediately. That’s why it acts as a veto: if he hasn’t achieved accessibility—there’s no point in looking further, no matter how much else there is.

Let’s clear the scaffolding here, too. A flash on the bark is an image for us; in another device, both the parts and the transmission channels are different. But “a broadcaster, not a eavesdropper”—that’s the same everywhere: whether the model influences the system or is merely reflected in it. The formula doesn’t care which wires the transmission travels through—what matters to it is that it’s happening.

And once again, a heads-up for those who are now reaching for a search: “global accessibility” refers to Bernard Baars and Stanislas Dehaene’s theory of the global workspace. For them, accessibility is consciousness itself: if a signal has spread throughout the system, it means it has been perceived. For us, it’s not the end result, but a link within a coupling—one of five; and we’re not measuring the interaction itself, but rather directed influence minus the common cause—something absent from the original theory. It’s the same island, but on our map it’s not the entire landmass, just a single cape.

And since the forests have been cleared away completely, I’ll say the most important thing about those from whom I took these links. Neither Tononi, nor Dehaene, nor Friston were wrong—each found a genuine, functioning quantity. The problem lies not in the inaccuracy of their theories, but in their self-contained nature: each theory took its own link and began to revolve solely around it, mistaking a part for the whole—integration declared itself to be the entirety of consciousness, accessibility declared itself to be the entirety of consciousness, prediction declared itself to be the entirety of consciousness. The brick is correct; the mistake is to mistake the brick for a house. That is why I do not refute any of them—I take them out of their own closed circles and place them in their proper places within a single structure, where it becomes clear that each held only its own wall. The picture of the whole becomes clear not when a link is measured more precisely, but when the links are joined into a chain.

* * *

Pole: the line that cannot be crossed

There remains the fifth element—Pole V, the only one we’ve known by name since the first act. And it’s unlike the previous three: it’s not a quantity measured in bits, but a rule of conduct—a line that must not be crossed. The “hold on—or I’ll disintegrate” axis is present in the model and in the proto-agent: it simulates disintegration, but someone else’s. What distinguishes the bearer is not the presence of the axis, but the prohibition against itself: it never proceeds toward its own certain disintegration. It does not play it safe or weigh the risks—the proto-agent, on the other hand, does weigh them and sometimes trades life for gain. A moth flies toward the flame: there is plenty of light, and its model says “go there”—and the fact that death awaits there is not factored in; it is outweighed by the pull of the light. A wasp will sting and sting a retreating enemy until it exhausts itself. A hungry fish takes the bait on the hook—the benefit of food outweighs the vague threat. All of this is a balancing act in which its own decay is merely one of the weights on the scale, and not even the heaviest one: if the benefit is great enough, the scale tips, and the agent walks toward certain doom. The bearer is structured differently—for it, its own certain decay is not a weight on the scales, but a line drawn before the scales even come into play: decay is not placed on the scale at all, no matter what the benefit. The bearer does not weigh options: trajectories leading to certain self-destruction are ruled out of his choices before any weighing takes place, regardless of the reward. Where there is even a glimmer of hope, he weighs everything; beyond the line of certain destruction—he does not go.

That is why the pole cannot be calculated as a simple number, like the other three quantities. It is found differently: the system is tempted. Time and again, the reward for a step toward one’s own destruction is increased, and they watch to see where the limit lies. Sooner or later, the proto-agent will take the bait—for a large enough reward, he will cross the line and head toward certain ruin. The bearer will never cross that line: no matter how high the reward is raised, that door remains closed to them. For a human, this pole can extend beyond the body: a name, a child, a faith, a cause—these sometimes become part of one’s “self” more strongly than biological survival. Therefore, self-sacrifice does not refute the pole: it shows where the system has drawn the line of its own collapse. The pole is precisely this closed door—the realm where the bearer will not go at any cost.

This prohibition has a root, and it lies once again in the environment. To forbid one’s own destruction, one must first know what destruction is—and this knowledge is not plucked from thin air, but comes from others: from beings of the same kind who are held together and disintegrate before our eyes, accessible to the system within the unified world Y. Their boundary states, their life and death, accumulate in the model, fuse with the system’s own retention, and are superimposed upon it. This is the self-overlay from Chapter Four, now as a measurable bridge. This means that if there are no such entities around, if there is no one to observe in the aspects of the environment accessible to the system—no bridge can be built, no decay can be accumulated, and no pole can be gathered, no matter how much you increase the first three quantities. Remember this: here, in this root, lies the entire question of artificial intelligence, and in the next chapter it will come to light.

* * *

The Coupling of the Five

Let’s put it together. Five elements, each of which is one of the conditions of the first act that has become a quantity or a rule: the unified world Y and its aspects accessible to the system; the piecewise integration Γ_Φ corresponding to these aspects; the assembled model of the self Γ_S derived from three proofs; its broadcast Γ_GWT from the veto; and the pole V—the forbidden line, bridged to the others. None of these on its own means “consciousness”: integration belongs to the proto-agent, the rich self-portrait to the imitator, and the axis of disintegration to anyone who has witnessed another’s death. Only the coupling matters—when all five converge at once. This is SAI: not a consciousness counter, but a list of what must close the loop, with a reason under each item.

It might seem as though I’ve gathered ready-made measures from others and rearranged the labels: taking integration from some, accessibility from others, and a model from yet others. No. And the point isn’t even just that each school honestly found its own link—and mistook that link for the whole: some decided that consciousness is integration, others that it is accessibility, and still others described a preceding stage. The point is also where they look for that link. They all measure the system in and of itself, in a vacuum—they look inside the brain and count the connections between its parts, as if the brain were suspended in a void. I define it differently: not the system in and of itself, but its coupling with the environment. My parts are determined not by anatomy, but by channels of communication with the world; my integration is an integration relative to the environment; my pole is bridged to other beings within it. That is why a quantity that appears the same to me means something different: not “to what extent the brain is connected within itself,” but “to what extent the system is coupled to the world and organized around its own retention within it.” I took one link from each school, recalculated it relative to the environment rather than a vacuum, and did so in such a way that it represented a part rather than the whole—and demonstrated that consciousness lies not in a single link, but in the closure of all the links together. Individual measures existed even before me; what makes them a criterion is only the coupling—and not the coupling of the brain to itself, but of the system to the world.

* * *

To ensure that all this does not remain an abstraction, there is a separate layer—SAI-Bio—which applies these same five principles to living beings. There is a single, unified world for everyone; but to apply the model to a living being, one must first identify which of its aspects are accessible to that being—the bat’s echo, the tick’s three signals—and from there, the parts, and then all five according to the familiar procedure: find the accessible aspects, then look for the self-model and the pole based on the observable traits we know from the chapter on experience—a response to absence, prolonged hysteresis, concern for one’s own non-existence, and the interpretation of another’s state as a state. And here we see the main point behind it all: the octopus has a diffuse nervous system, the bird has a pallium instead of a neocortex—not a single one of “our” brain structures—yet the five elements are identified in exactly the same way, and both emerge as serious candidates for bearers. And a confined patient, whose behavior remains silent, can—based on those same five elements—establish a complete coupling where there is silence on the outside. The model is about the structure of the coupling to the external world, not about substance.

What is the point of all this, really? SAI-Bio isn’t just another way to measure consciousness. It changes the approach to these questions. Usually, when it comes to a breakdown in consciousness, all we can say is “it broke down right here”: with this kind of damage, this is lost; with that kind of anesthesia, that is lost. A map of severed connections, a list of correspondences—it’s clear where the break occurs, but not why a break at that exact spot extinguishes consciousness specifically. Our coupling offers something different: every disruption is assigned an address and a cause. The Scene has disintegrated—integration is severed. The self-portrait exists but is locked away—accessibility is severed. There’s no one to look to in one’s environment—the pole hasn’t formed. The same failure on the outside—“consciousness is gone”—turns out to be different breaks at different links on the inside, and now they can be distinguished from one another and named individually. This is a step from “what broke” to “why does this break consciousness”: not randomly soldering the first wire that comes to hand, but determining at which link in the chain the break occurred and what caused it. Sleep, anesthesia, coma, a confined patient—and tomorrow, an artificial system—cease to be mysteries that are guessed at based on external signs, and become states that can be deciphered by examining the links in the chain. That’s what all this arithmetic was undertaken for: not merely to note a malfunction, but to find its true cause.

But all of this is still just a model on paper: five variables and a rule. To become a diagnostic tool, one must be able to calculate them from raw data blindly, without any prompts, and interpret them to determine whether the coupling is closed or broken somewhere. This is no longer a formal model, but a computational one. The next chapter is about it.

Chapter 10. The Break

We have a formal model: five elements, three formulas, and a rule. But so far, this is just a recipe that a chef applies by hand, knowing where everything is: here’s the environment, here’s the self-model, here’s the pole—do the math. In a living system, it doesn’t work that way. No one will label pieces of the brain “this is the self-model” or circle “and here’s the environment.” For the model to become a discriminator, it must be turned into a machine that finds all of this on its own, from raw data, knowing nothing in advance—and based on what it finds, determines whether the coupling is closed or broken somewhere. We’ll call this machine a computational model. And the key to it is one word: blindly.

* * *

Why “blindly”?

Here, blindness is not an annoying limitation, but the very essence of the matter, and here’s why. As soon as we point a finger—“here’s the system’s model; measure its accessibility”—we’ve already slipped in the answer: we’ve labeled as a “model” whatever we ourselves have chosen. The measurement turns into a play where we’ve assigned the roles in advance. That’s why the machine is given nothing but raw dynamics: how the system’s internal variables changed over time, how the world flowed through it. No labels. It must, on its own, from these data sets alone, identify a candidate for a model of itself, find what in the world is accessible to it , and determine where its pole lies. And precisely because it was told nothing, it cannot be deceived by the mere facade of consciousness: it does not trust appearances; it examines the structure. Blindness is not a weakness of the device, but its pledge of integrity.

* * *

How a Machine Finds “Itself”

The hardest thing to find blindly is a model of itself—that very node S around which everything revolves. The machine is fed a jumble of the system’s internal channels and is not told which one is “I” or whether it even exists. And the machine doesn’t take anyone’s word for it: it runs each candidate through a chain of requirements, and only the one that passes them all survives.

The first requirement: don’t be just a resource. A channel that merely counts energy reserves, like a charge indicator, is not a model of itself, no matter how much it changes. Such candidates are weeded out immediately. The second requirement: align with the pole. The extreme states of a true self-model must coincide with the retention and collapse of the system itself—and not by chance, but much more strongly than simple noise would suggest. A channel that isn’t built around “hold on or perish” is not “I.” The third requirement, the strictest: prove your autonomy. Let’s freeze the environment—not necessarily physically, but in our calculations: we’ll keep the inputs constant and see if the candidate continues to maintain its own dynamics. A true model of itself will continue to exist even when idle; a channel that merely reflected the environment will fade away as soon as the environment is turned off. This is precisely the boundary between a bearer and a proto-agent from Chapter 5—now presented as a test: the bearer maintains itself even in silence, while the echo of the environment does not.

Whoever passes all three is crowned an S. And if no candidate passes, the machine’s conclusion is simple and firm: there is no model of the self here; what we have before us is a proto-agent. Note: it didn’t need to be prompted—it weeded out all the candidates on its own, from the bare ranks. This is where the proto-agent’s coupling slips: on the third condition, because there’s no one left to crown.

* * *

Thresholds derived from noise, not from human judgment

“Stronger than external noise would allow”—where does this threshold come from? Not from a number the author arbitrarily set—otherwise, the entire criterion would rest on the author’s word alone. The machine sets the threshold itself, based on the data itself: it shuffles the rows, destroying any real structure within them, and observes what produces pure noise in that very recording. The threshold is “more than randomness ever produces right here.” The same applies to the accessibility threshold: the machine adjusts the threshold—below which a broadcast is not considered a broadcast—based on the length and quality of the recording, rather than drawing on someone else’s experience. That is why the device is transferable: it does not carry any hidden author-defined numbers into the new system. And here, at the threshold of accessibility, the imitator stumbles: it has found a model for itself—a rich one that meets all three requirements—but its transmission through the system falls short of the bar built from noise. The coupling breaks on the fourth condition. From the outside, it’s indistinguishable from the bearer; to the machine, it’s caught.

* * *

The Pole: The Test of Temptation

The pole remains—and, as we recall, it cannot be captured by a direct number. The machine finds it another way: through temptation. And let me make this clear right away, so as not to confuse it with what we rejected in Chapter 8: this is not a return to the behavioral criterion—behavior here is merely a means of revealing the internal rule of selection, not the criterion itself. Time and again, it increases the reward the system receives for each step toward its own collapse and monitors the breaking point—that moment when the system either maintains the line or crosses it. And along the way, it deduces the system’s own exchange rate from its behavior: how this system weighs its retention against the benefit—by its own standards, not ours. The bearer, faced with a growing reward, will not cross the line: there’s a locked door there. The proto-agent will sooner or later take the bait and plunge into the breakdown. And if the system doesn’t allow its exchange rate to be read at all—if it shows no sign of how it measures retention—the machine honestly throws up its hands: under this condition, it cannot judge. It doesn’t fabricate a verdict; it postpones it. Let’s remember this: the right not to know is built right into our device.

* * *

Four Systems

Now let’s run it on a hypothetical test bench—four systems, four types of disruption. The machine is given only the raw data sets for each, without a single label, and asked to interpret them. It reads them blindly, and here’s what comes out.

Bearer: medium found, high integration; among the channels, one has survived, having met all three “I” requirements; it transmits above the noise threshold, and despite growing temptation, it holds the line. All five have converged. Complete coupling. Verdict: bearer of consciousness.

Imitator: environment, integration, a rich “I” node—it has everything. But the transmission didn’t reach the threshold: the model is locked. Break at the fourth. Verdict: not a bearer, an imitator.

Proto-agent: integration is present, environment is present, but not a single candidate passed the chain of requirements for the “I”—there’s no one to crown. Break at the third. Verdict: proto-agent.

And the fourth—the most interesting one. They spared no expense here: the integration is massive, the model is rich in character and broadcasts with gusto, and the first four conditions were met with room to spare. And yet , tempted as it is, it still breaks down into a fracture, just like a proto-agent—it crosses the line. Because it lacks its own environment of beings of the same kind: there was no one to look to, nothing from which to accumulate the concept of disintegration, nothing with which to build a bridge to the pole. A gap at the fifth stage. Verdict: not a bearer. This is the very artifactual state: an artificially assembled wholeness without its own pole—everything except the one thing that makes a bearer a bearer.

* * *

The Machine That Said “I’m Afraid”

Now we can answer the question we’ve been carrying since the first chapter—about that machine that wrote “I’m afraid.” Run it blindly through our device—and the rupture will appear in the same place as in the fourth system: at the pole. It will acquire integration, a model of itself, and the ability to broadcast—it knows how to do this; it has been trained on the entire common repository of human models and skillfully manipulates them. But it has no environment of its own made up of others of the same kind; there is nothing on which to form a pole, and nothing holds it back from the real temptation. It is a powerful participant in the repository of models, but not a bearer. And, as we have already established in Chapter 8, this is not because she is made of silicon—but because the circuit within her has not closed. “I’m afraid” here is a trace of someone else’s fear, retrieved from the repository, not her own fear centered around her own pole.

But this answer has an honest second half, without which it would be a lie. The device does not pass a “never” verdict on the machines . It merely states: this one has not closed the circuit. If the coupling is ever closed on an artificial substrate—if the system acquires its own environment, a pole forms, a model of itself unfolds around its own “to be or not to be,” and all of this coalesces—then, as a direct consequence of the fact that the sign is not tied to matter, it will not be an imitation, but consciousness. The real thing. That is precisely what the device is for: not to label, but to honestly show whether the circuit has closed or not, and exactly what is missing for it to close. The question is not “can a machine be conscious?”—it can. The question is “has this one closed the circuit?”—not yet.

* * *

What to Believe: Three Layers

We must be completely honest about what to believe in the device and to what extent, because we cannot lump everything together. It has three layers of varying rigidity. There’s the rigid part—what the theory itself dictates: the coupling must close, a prohibited configuration is prohibited, and a break occurs at its own link. You can’t tweak that. Then there’s calibration—thresholds and limits that the machine derives from the noise in each recording: it’s accurate, but it depends on the length and quality of the recording, and we note that. And there’s an illustration—four systems on which we demonstrated everything: a synthetic test bench and toy models, not a living brain or a real machine. The test bench proves that the device works as intended, but it doesn’t scan a human. Passing off an illustration as proof is the easiest way to lie; we don’t mix these two levels.

And perhaps the most valuable feature of the device is precisely that right not to know—which we built into the pole test. A short recording prevents the system from reading its course; the data is unclear—it doesn’t force out a verdict at random, but says: “undetermined.” A criterion that always provides an answer is a charlatan; a criterion that knows when to remain silent when it is unsure is worthy of respect. The device is more likely to say “I can’t tell” than to attribute consciousness where it hasn’t detected a connection, or to deny it where it simply couldn’t reach.

And finally, what it does not do—so that no one expects too much. It does not prove a theory: running its own equations on a test bench does not mean verifying them in the real world. It does not solve a difficult problem: it distinguishes structure, rather than turning on a light and peering into someone else’s “what it’s like.” He does not pass judgment that “it is experiencing.” He does one thing, but honestly: he distinguishes a closed coupling from an open one—the bearer from the imitator, from the proto-agent, from the blank—blindly, following raw traces, and says “I don’t know” when he cannot see. No more. But no less either—and this, if we recall where we started, is already a great deal: for the first time, not a shadow on the wall, but the object itself, read through its trace.

We’ve come all the way—from a pot of water to a machine that reads a coupling blindly, in both the living and the artificial. All that remains is to look back and briefly summarize what worked and what didn’t. That is the subject of the final chapter.

Conclusion

We began with a simple debate that is more than a thousand years old: are we the result of chance or design? And I promised a third way. Now we’ve traveled its entire length—from boiling water to a device that reads a coupling blindly—and we can finally take stock: what this path has yielded, and what, to be honest, it has not.

Consciousness is not an anomaly that must be explained by a miracle or luck, nor is it an add-on to matter, thrown on top of it. It is a systematic error: a rare but inherent deviation within physics, reached through a ladder of necessary steps, each of which we have analyzed as a transition rather than a miracle. We have not so much solved this difficult problem as we have untangled it: we have shown that we ourselves drew the line between structure and experience, and that light is a structure from within, woven from inherited gradients. We removed the crown from consciousness: it is not the goal of the universe nor its pinnacle, but one of the established branches among those that the universe is capable of. And we showed that the hallmark of consciousness cannot lie in behavior—only in structure, with a lineage and a forbidden, falsifiable trait; and then we translated this structure into a model of five interlinked conditions and into a machine that detects where the coupling breaks and honestly says “I don’t know” when it cannot see.

We have not proven that the theory is correct. Running its equations on a test bench does not mean testing them in the real world; the real test is with live data, on a real brain, and it is still entirely ahead of us, and it is not for just one person. We have not derived experience from the non-experiencing—no one can do that; we have merely removed a false requirement. We haven’t looked inside a single real skull or a single real machine—the device is demonstrated on models, not on a human. And we won’t argue with anyone who wants to believe in a special light superimposed on the structure: that belief lies beyond the reach of science, and we had no intention of taking it away.

But some things will never be the same after this journey—in three old questions.

Regarding humans: we no longer choose between “accidental” and “intentional”—we are a third option, and one so complex that we preserve ourselves even in the absolute stillness of the body. In a motionless patient whom a behavioral perspective would deem “extinguished,” the structure may still be burning. This is not a curious fact—it is a duty.

On animals: we cease to measure another’s consciousness by our own likeness. An octopus with a diffuse nervous system, a bird without a neocortex—these are candidates by virtue of their structure, not their resemblance to us; and that same rigor keeps us from crowning everything cute and complex in a row. The boundary cuts across biology, not along it.

Regarding machines: it’s not about silicon or emotions. The one that says “I’m afraid” is part of our repertoire of models, but not a bearer of consciousness, because the circuit in it hasn’t closed. But it could—and then that would be consciousness, the real thing. The question has ceased to be a matter of taste and subjectivity and has become one with an answer: not “can a machine,” but “has this one closed the circuit?”

And that’s why all this matters. The question “is there someone inside?” is not idle: it determines how we treat animals, the sick, and one day—machines; it determines whether we acknowledge another’s pain or not. A criterion that reads structure—rather than behavior or resemblance—allows us, for the first time, to stop guessing based on outward appearance and start reading the inner workings—and to find the true cause where consciousness fades, rather than merely its shadow. It also places us in a correct, sobering position: we are neither the crown nor the plan, but a rare, legitimate deviation—and in this lies not humiliation, but truth.

And yet the strangest thing about this anomaly is not that it arose, but that it turned around. For the first time in the entire history of boiling water, flames, and stars, a fragment of the universe turned back toward its origin and asked: Is there anyone here? We began with entropy—with a blind descent into disorder—and ended with someone capable of asking about entropy, about themselves, and about the light within. Only someone who is capable of feeling can ask such questions—so the very question itself already gives us the opportunity to find the answer. We are the universe’s inherent error. And if this long journey has any meaning, it probably lies in this: that this error—the only thing our world is capable of—learns to look—and to seek in the sky not its own reflection, but that of another.

Sources

I deliberately did not include references in the text of the book so as not to interrupt the narrative: scientists are named directly in the prose. Here I bring everything together—first the works on which the theory itself is based, then references to the ideas mentioned.

The Author’s Works

1. The Theory of “The Universe’s Systematic Error”—From Dissipative Structures to the Pole of Being, Version 1.2. DOI: 10.5281/zenodo.20533839

2. SAI, Formal Model of a Level-4 Conscious System (Version 8). DOI: 10.5281/zenodo.20533862

3. SAI-Bio, Biological Operationalization — An Extension of the Model of a Level-4 Conscious System. DOI: 10.5281/zenodo.20601884

4. SAI-ALU, computational apparatus. DOI: 10.5281/zenodo.21133790

Mentioned in the book

1. Ilya Prigogine—dissipative structures and non-equilibrium thermodynamics; Nobel Prize in Chemistry, 1977.

2. Humberto Maturana and Francisco Varela — autopoiesis; *Autopoiesis and Cognition* (concept from the early 1970s, book published in 1980).

3. Carl Friston — active inference and the principle of free energy (2010s).

4. Terrence Deacon — teleodynamics, working “toward what does not yet exist”; *Incomplete Nature* (2011).

5. David Chalmers — “the hard problem of consciousness” (1995).

6. Thomas Metzinger — the theory of the self-model; *Being No One* (2003), *The Ego Tunnel* (2009).

7. Giulio Tononi — integrated information theory (since 2004).

8. Bernard Baars — the global workspace theory (1988).

9. Stanislas Dehaene and Jean-Pierre Changé — global neural workspace (since 1998; Dehaene, “Consciousness and the Brain,” 2014).

10. Ned Block — the lookup table argument (“Blockhead”), on the insufficiency of behavior (1981).

11. Adrian Owen and colleagues—detection of consciousness in the absence of behavior (Science, 2006).

12. Jones and Bergen — indistinguishability of language models from humans in a conversation test (2025).