Everett's Demon

Ontological randomness and the amplified quantum

Whenever I claim that a chess game is genuinely open — that the outcome is not fixed until the game is played — the same objection arrives on schedule. Surely, the critic says, this randomness is merely epistemic. Grant a hypothetical intelligence perfect knowledge of every neural state and every particle position in both players’ brains, and it would call every move before it was made. The uncertainty you feel at the board is just ignorance dressed up as metaphysics.

The demon in that objection has a name. Pierre-Simon Laplace imagined an intellect that, knowing the position and momentum of every particle and the laws governing them, would hold the entire future in a single glance: “nothing would be uncertain, and the future, as the past, would be present to its eyes.” Laplace’s demon is classical determinism personified — one set of initial conditions, one set of laws, one future. On that physics, all unpredictability is a confession of incomplete information or insufficient computing power, and the critic wins: measure the players finely enough and the game is a foregone conclusion.

But that physics is false, and it has been false for a century. The objection does not survive the upgrade to modern physics.

Two Demons

Quantum mechanics establishes, experimentally and independent of interpretation, that the outcome of such an event cannot be read off any accessible prior state: the formalism assigns amplitudes to alternatives rather than a unique result, and Bell tests rule out saving the appearances with local hidden variables. Whether a particular ion channel opens, whether a synaptic vesicle releases its neurotransmitters, how a molecular binding resolves — these are not outcomes a sharper microscope would reveal in advance. What that unpredictability is — irreducible chance, every alternative realized, or a determinate order carried in nonlocal variables — is exactly where interpretations divide, and the formalism alone does not settle it. This book takes the Quantum Branching Universe (QBU), on which those alternatives do not stay merely possible: as decoherence renders them dynamically independent, they settle into distinct branches — approximate, emergent histories at the relevant grain rather than a bookkeeper’s exact per-event forks, but real, and irreversibly diverging.

So let us grant the critic everything and see what it buys. Take the demon seriously and give it complete knowledge — not the classical fiction of exact positions and momenta, which quantum mechanics forbids, but the real thing: the universal wavefunction, every superposition, every entanglement, every branching point. Call this upgraded intellect Everett’s demon, after Hugh Everett, who first took the wavefunction’s own dynamics at face value.

Everett’s demon is not a weaker Laplace’s demon. It knows everything there is to know, and the laws it applies are exactly deterministic — the wavefunction evolves under the Schrödinger equation without a flicker of chance. Yet what it foresees is not a single future. It foresees the branching structure whole: every possible line of the game, each played out in a physically real timeline, each carrying its objective weight of Measure. Ask it who wins tomorrow’s match and it does not name a winner. It hands you the tree.

That is the answer to the critic, and it is worth stating sharply: perfect knowledge does not restore the single future. Perfect knowledge reveals the plurality. The critic’s thought experiment, run on correct physics, proves the opposite of what it was built to prove.

One Distinction

The objection trades on a conflation that is worth dissolving exactly once, because everything downstream of this chapter depends on it: the conflation of determinism with a closed future.

Consider the digits of π. They are as determined as anything can be — fixed by a one-line definition, immutable, identical in every branch of every timeline. Yet they behave statistically like a random sequence (π is believed, though not proven, to be a normal number), and predicting the trillionth digit without computing your way there is hopeless. This is epistemic randomness in its purest form: nothing about the world is open; only your access to it is limited. There is exactly one fact about the trillionth digit, and your uncertainty measures the distance between you and it.

Now consider the future of the branch you occupy. The QBU is also globally deterministic — the demon’s-eye view has no dice in it. But you are not the demon. You are an agent embedded at a Vantage inside the structure, and forward of your Vantage the timelines genuinely diverge: decoherence makes the splits irreversible, multiple incompatible outcomes are all physically realized, and no possible improvement in your knowledge tells you “which branch you will end up in,” because that question has no single answer — you end up in all of them. This is ontological openness: the uncertainty is not a gap between you and a fact; it is the structure of the facts themselves. The digits of π are closed and merely hard to reach. The future is open, full stop.

Both systems are deterministic from the global view; only one of them leaves the local future open. That is why “but the wavefunction evolves deterministically!” is no comfort to the critic — determinism at the level of the whole structure is compatible with, indeed produces, irreducible openness at every Vantage within it. The full taxonomy of uncertainty, and the discipline of keeping Credence pointed at the right kind, belongs to Varieties of Uncertainty; here the point is physical: some unpredictability is ignorance, and some is branching, and no demon converts the second kind into the first.

The Amplifier

The critic has a fallback: quantum events are microscopic, and brains are warm, wet, and macroscopic. Doesn’t all that indeterminacy average out into effectively classical neural machinery?

It would, in a system that damped it out. Brains, this chapter conjectures, may do the opposite. Neural signaling runs on components that operate at scales where quantum fluctuations can matter: probabilistic neurotransmitter release, single ion channels gating open or closed, molecular bindings resolving one way or the other. A neuron sitting near its firing threshold is a plausible detector of exactly this noise — a sub-threshold fluctuation could decide whether it spikes now, a few milliseconds later, or not at all. Feed that spike into a recurrent network that discriminates, decides, and acts on fine differences, and the microscopic variation need not wash out; it can cascade.

There is a sharper form of the fallback, and it is the one that has to be met. It is not that indeterminacy averages out but that it decoheres. Max Tegmark’s estimate is the standard statement: a superposition of a neuron’s firing-versus-resting states decoheres in roughly \(10^{-20}\)–\(10^{-19}\) seconds, and even the microtubule excitations of the most optimistic proposals in about \(10^{-13}\) — against neural dynamics measured in milliseconds, a separation of many orders of magnitude.¹ Long before a superposition could persist to steer a spike, the environment has decohered it. If the brain needed to hold quantum states — to compute with them, keeping alternatives interfering until a later operation read them out — this would be decisive, and the proposals that require it, such as a coherent quantum computer in the microtubules, do not survive it.

But that is not the claim here. Amplification does not need coherence to survive; it needs an outcome that is open. Decoherence is precisely what differentiates a superposition — at the relevant grain — into effectively autonomous, non-interfering components, all of them realized, that can differ in the decisive coordinate; the mechanism wants those components differentiated fast and then amplified, not kept interfering. So Tegmark’s timescale refutes quantum computation in the brain, not the amplification of quantum indeterminacy. What it does not hand us for free is the step that matters: decoherence secures only that the components are well-defined and non-interfering; the amplifier needs in addition that they differ where it counts — that the spread of the decisive collective coordinate across those components straddles the firing threshold with non-negligible Measure. That is a quantitative conjecture about how finely a neuron is poised, and it is one Tegmark’s calculation does not address: he computes how fast a superposition of firing-versus-not-firing states decoheres, not whether the decohered components differ across threshold to begin with. His own conclusion — that cognitive dynamics can be modeled classically and stochastically — is compatible with either answer. That the biologically decisive components straddle threshold is therefore a further, and presently unsupported, conjecture, which this section owns rather than borrows.

A distinction underneath it has to be kept, too. The differentiation into components is the quantum step; the millisecond of amplification that follows runs within a component, where the dynamics are quasi-classical and effectively deterministic. A later Vantage conditions on one definite record, but nothing selects a branch — every component is realized. The amplifier does not run on a live superposition; it enlarges a difference already fixed in the decoherent record. So the only question is whether that differentiation is genuinely plural — all components real — or merely epistemic: whether the fluctuation is quantum indeterminacy or ordinary thermal noise, unpredictable only because we cannot track every molecule. Here the book’s ontology answers, with a flag. Assuming a universal Everettian state and a chosen decoherent coarse-graining, a local thermal state is not a proper mixture — ignorance over a hidden definite microstate — but an improper one, the reduced shadow of the global superposition. Thermal stochasticity can then be represented as branch-relative quantum dynamics rather than classical ignorance; it need not be read as a hidden definite microstate. (Whether any single fluctuation counts as one discrete branching event is a matter of the approximate, coarse-graining-dependent branch structure, not a sharp fact.) But this is a consistency claim within the Everettian picture, not interpretation-neutral evidence: a Bohmian would locate a definite hidden microstate and read the same openness as epistemic. Granting the QBU, the seed need not be certified quantum-rather-than-thermal; deny it and the objection returns. The dissolution settles the character of the seed, not the magnitude question above.

This buys the openness — but not that brains are the special place where the microscopic becomes historical. A rock branches as much as anything; near equilibrium its components stay macroscopically indistinguishable, their differences never amplifying into a durable coarse-grained record. Branching is ubiquitous; amplification is not. Any system poised near an instability amplifies quantum indeterminacy into macroscopic difference: a Geiger counter turns one undecided decay into an audible click, a weather system turns molecular jitter into which day it rains. That is the first of two conditions, and it is generic — storms and counters meet it.

What they lack is the second: the amplified outcome routed through a control loop — a system whose later states track its policy rather than drift, agency as Part I defined it. The Geiger counter’s click is a dead end; the storm answers to no policy and is weighed by nothing. Only where the open outcome is taken up — perceived, weighed against the system’s remembered ends, and issued as an action — does it become a difference in behavior the system can be said to author. That uptake comes in degrees: a thermostat wired to a noisy sensor is its thinnest instance, a plant’s tropisms take up more, a deliberating animal is the thick end where “authored” earns its weight. Only this second condition distinguishes an agent from a thundercloud; the first is raw material both share.

One honesty is owed, and one payoff claimed. The control loop is substrate-neutral: nothing in the uptake requires the seed to be quantum rather than a good classical pseudo-random source, and an agent built on the latter would behave identically. So quantum indeterminacy does exactly one job here — not the source of agency, which is substrate-neutral and belongs to the uptake, but the difference between an agent whose unchosen alternatives were never real possibilities at all, only ignorance, and one for whom the road not taken is an existing branch. Whether that difference matters is the reader’s to weigh; it is the only thing the quantum ontology is here asked to secure — and it is secured for the rock’s future as much as the agent’s. Ontological openness is universal; only its uptake into authored action is special. A critic indifferent to whether the openness is real or merely unknown can take the agency and leave the quantum ontology on the table, and nothing here stops him.

Where in cognition does this actually bite? Not uniformly; it concentrates where decisions are close — a comparison of two nearly equal options, a wavering of attention, the thinned margins of fatigue — the points at which a branch-relative difference in a single near-threshold event can tip which way the whole system goes. An agent’s cognition is, in this respect, a chaotic system with an open seed. Classical chaos is deterministic sensitivity: one trajectory, exquisitely dependent on initial conditions, unpredictable only because you cannot measure finely enough. Here the sensitive dependence is fed by quantum differentiation whose components had no single outcome, so the result is not one hard-to-predict trajectory but real branching: physically distinct futures, all realized, diverging at the speed of thought.

The Hard Case and the Easy Ones

Chess is the hard case, which is why I lead with it. No dice, no shuffled deck, no bounce of a ball — the game is widely taken as the epitome of deterministic cognition, pure strategy executed by pure intellect. If quantum branching reaches even here, it reaches everywhere.

It reaches here. Every evaluation, visualization, and calculation in a chess player’s head is neural activity, and the branching enters through the concentration points just named. Move selection: a player weighing two candidate moves of nearly equal strength is running a comparison whose outcome sits within the noise band — branches diverge, and in some timelines she plays the bishop, in others the knight, and the games that follow have nothing in common. Attention: whether she spots the quiet tactical threat on the queenside is a matter of where a quantum-jittered spotlight lands, and spotting it or missing it is the difference between winning and losing. Calculation: a fatigue-loosened slip in working memory, quantum-seeded, turns a sound combination into a blunder in this branch and not in that one. From the Vantage of the first move, a chess game is not a hidden fact awaiting discovery. It is a Branchcone: the weighted ensemble of games played forward from here.

Once the hard case falls, the easy ones follow a fortiori. A hockey game is millions of micro-decisions under time pressure — shoot or pass, the angle of the stick, the goalie’s read — each one a near-threshold neural event, each fraction-of-a-second variation in reaction time capable of converting a save into a goal and rewriting the rest of the game. A conversation branches at every word choice, every flicker of what to say and when. Whether you step off the curb this instant or the next is quantum-shaded. None of this unpredictability is mere complexity, the epistemic fog of too many variables. It is ontological: daily life is a continuously branching structure, and the openness you feel about how the game, the negotiation, or the evening will go is the accurate perception of a future that is genuinely plural.

And the a fortiori runs upward as well as downward. If the most formalized of intellectual activities branches, then the less formalized ones — scientific discovery, technological invention, political conflict — branch more, not less. The open future is not a special property of games. Games are just where it is easiest to see.

Candyland

To see what the absence of openness looks like, take the perfect contrast case: Candyland. Children draw cards from a shuffled deck and move their pieces where the cards dictate. There is suspense, cheering, heartbreak — and not a single decision. The outcome of a Candyland game is completely fixed at the moment the deck is shuffled. Everything after the shuffle is transcription. A player’s uncertainty about who will win is purely epistemic, exactly like uncertainty about a distant digit of π: know the order of the deck and the game is over before it starts, in every branch alike.

Candyland is a working model of the Laplacean universe — a world where the future is closed and all suspense is ignorance. Chess is a working model of the Everettian one. Same table, same children, same excitement; ontologically, different kinds of world. The difference is agent choice. Both children’s brains are branching all evening — Candyland simply gives that branching nothing to do, because the deck settled the game before a decision was possible. In chess the player’s branching is routed through the moves she chooses, so which game gets played depends on which way she goes. Agents are not the mechanism by which microscopic openness becomes macroscopic — storms and Geiger counters manage that. They are the mechanism by which an open future becomes an authored one.

Absurdity and Freedom

The contrast is not only physical. Which world you are in — Candyland or chess — is the hinge of an old quarrel about whether a life can matter.

Camus built his account of the absurd on the image of enacting a script that is already written — Sisyphus at his boulder, investing effort and passion in an outcome fixed in advance. Candyland is absurdity in miniature: players construct narratives, feel hope and despair, celebrate victories, all around a result that was settled at the shuffle. Nothing they do makes any difference to anything, and the emotional life of the game is a story told about a foregone conclusion.

Sartre’s freedom is the other picture: the agent confronting a future that is genuinely open and bearing the responsibility of shaping it. Chess, hockey, conversation — every activity routed through choosing brains — is on this side of the line, and not as a metaphor. The openness Sartre demanded is physically real: your decisions are branching events, the futures you weigh are timelines you are about to distribute yourself across, and shaping the open future is a causal description of what a deciding brain does to the branching structure.

Laplace’s universe made Camus right about everything — a cosmos of Candyland, all suspense and no stakes, where the deepest fact about your deliberation was that its conclusion predated it. Everett’s universe puts most of human life on Sartre’s side of the line. The demon with perfect knowledge does not see through your choice to its predetermined outcome; it sees your choice doing what choices do — splitting the world. Whether that splitting deserves the name free will, and what freedom amounts to when every option gets taken somewhere, is the question of the next chapter.