Axio › Volume 1 › The Interpretation Wars

The Interpretation Wars

QBism, local realism, and what the QBU keeps

This chapter is a draft — it is readable but still changing.

Quantum mechanics is the most precisely confirmed theory in the history of science, and after a century physicists still cannot agree on what it says. Two of the live contenders could hardly disagree more. QBism — Quantum Bayesianism — holds that the quantum state is not a description of the world at all: it is a catalogue of one agent’s personal beliefs, and “collapse” is nothing but that agent updating on new evidence. The many-worlds tradition holds the opposite: the wavefunction is fully objective, universal, and real, a deterministic structure that branches at every quantum event. One side says the state is in your head; the other says it is everything there is. And no experiment ever performed, or ever performable, can tell them apart.

That last fact is what makes this a war of interpretation rather than a scientific dispute in the ordinary sense. But a war without empirical referees is not a war without winners. The choice among interpretations is philosophical, and philosophical choices can be made well or badly. This chapter fights the Quantum Branching Universe’s corner (The Quantum Branching Universe gives the construction) against its two most instructive rivals: QBism, whose central insight the QBU absorbs while refusing its central deflation, and the local-realism challenge, which turns out — through a remarkable result of Paul Raymond-Robichaud’s — not to be a rival at all but a mathematical vindication.

Why Experiment Cannot Referee

QBism and many-worlds run on exactly the same machinery: the Schrödinger equation and the Born rule. They generate identical predictions for every conceivable experiment. Distinguishing them empirically would require one of three things: observing direct interaction between separate branches, which the branching picture itself rules out; detecting an objective collapse event, which both interpretations deny exists; or finding deviations from standard quantum predictions, which neither anticipates. The disagreement is not about what the world will do. It is about what the formalism is.

So the arguments that remain are the classical philosophical ones. Logical coherence: does the interpretation avoid contradiction and paradox? Parsimony: does it minimize assumptions — assumptions, not entities, a distinction that dissolves the tired charge of “ontological extravagance” against branching, as the starlight analogy in The Observer Joins the Branch shows. Explanatory clarity: does it make quantum phenomena natural rather than mysterious? These are real standards, and interpretations pass or fail them differently. Empirical equivalence does not make the choice arbitrary; it makes the choice philosophical.

What QBism Gets Right

QBism’s core move deserves to be taken seriously, because part of it is correct.

The QBist says: a quantum state assignment is a personal Bayesian probability — a credence. When an agent measures a system and “the wavefunction collapses,” nothing physical has snapped; the agent has acquired evidence and conditioned on it, exactly as a Bayesian conditions on any evidence. Quantum mechanics, on this reading, is not a mirror of nature but a user’s manual: a normative guide for belief revision, prediction, and decision under quantum uncertainty. The measurement problem, which has embarrassed physics since the 1920s, dissolves — there was never a mysterious physical process to explain, only an agent updating.

The insight buried in this is genuine: the probabilities an agent uses are subjective. They live in the agent, they differ between agents with different evidence, and they change by Bayesian updating rather than by physics. Any interpretation that ignores the agent’s side of quantum probability ends up smuggling it in somewhere. QBism put the scientist back into science, as Mermin’s slogan has it, and on this point it is simply right.

Where QBism Overreaches

The mistake is what QBism does next: having correctly identified the subjective layer, it declares that layer the whole story. There is, for the QBist, no observer-independent quantum state, no objective quantum probability, nothing that the agent’s credences are credences about beyond the agent’s own future experiences.

This deflation costs too much. If quantum states are only beliefs, there is no fact that makes one agent’s beliefs better calibrated than another’s; yet the entire practice of physics consists in some probability assignments being right — vindicated to eleven decimal places — and others being wrong. A theory of pure belief cannot explain its own predictive success, because success is a relation between belief and something that is not belief. And a science whose subject matter is each user’s private experience has quietly given up on being about a shared world.

The QBU keeps the insight and refuses the deflation, by refusing to make one probability concept do two jobs. Measure is the objective weight of a branch — a physical quantity, as real as mass, defined in Measure, Vantage, Branchcone; Credence is the subjective probability an agent assigns from its vantage, governed by the Bayesian epistemology of Measure and Credence. Credence tracks Measure when the agent is doing its job; the gap between them is exactly the gap between belief and world that QBism erased. Everything QBism explains, this two-level picture explains — collapse as updating, probability as credence, the agent at the center of the epistemic story — while keeping the objective ontology that makes the epistemic story worth telling. QBism is the QBU with its bottom layer amputated.

Relational quantum mechanics deflates in a different direction — making reality itself relative to interactions — and the previous chapter, The Observer Joins the Branch, already showed why observation is ordinary physical correlation rather than reality-making.

The Local-Realism Vindication

The second engagement comes from the opposite flank. QBism attacks the realism of the branching picture; a longer tradition attacks its locality. Bell’s theorem is routinely glossed as proving that no local-realistic account of quantum mechanics is possible — that entanglement forces a choice between locality and realism, and any interpretation claiming both is cheating somewhere.

Paul Raymond-Robichaud’s local-realistic model for quantum theory refutes that gloss constructively, and in doing so builds, within physics, the very architecture the QBU asserts philosophically.

His construction rests on a rigorous split between two levels of description. The noumenal state of a system is its complete ontic state — everything that is physically the case about it, whether or not any observation could reveal it. The phenomenal state is what observation can access: the measurable face of the system. The two are linked by an epimorphism — a surjective map from complete ontic structure down to the observer’s partial informational projection. Anyone who has followed this book will recognize the shape immediately: it is the Measure/Credence split, stated as mathematics. The noumenal level is where Measure lives; the phenomenal level is all any agent’s Credence can answer to. Both frameworks stand on the same refusal: observables do not exhaust reality. Where QBism deflates the noumenal level away, Raymond-Robichaud proves you can keep it — locally, and without contradiction.

The central theorem is an equivalence: a theory forbids action at a distance if and only if it forbids observable action at a distance. There is no observable nonlocality in quantum mechanics — entanglement famously cannot be used to signal — and therefore, by the theorem, there is no real nonlocality either. No hidden spooky action lurking beneath a no-signaling surface. What look like nonlocal correlations are informational coherence across branches — records of shared history — not causal influence between separated systems.

For this book, that theorem is a load-bearing beam. Local realism, in Axio’s language, is agency preservation: each agent acts within its own decoherent branch, and nothing any agent does reaches across space to violate the causal autonomy of another. If quantum mechanics were genuinely nonlocal, the picture of an embedded agent steering its own branchcone would be physics-adjacent poetry. Raymond-Robichaud’s result says it is just physics.

Projection, Not Foundation

The construction carries one more consequence, and it cuts against naive Everettianism as sharply as against QBism. Raymond-Robichaud proves that the universal wavefunction cannot serve as a complete noumenal description: it encodes the phenomenal correlations perfectly, but it omits the noumenal separability — the fact of the matter about individual local systems — that locality and individuation require.

This formalizes a critique I have pressed on other grounds. Treat the universal wavefunction as the totality of reality and you erase vantage: there is no longer any fact about where in the structure an agent is, and with vantage goes agency itself. The wavefunction is a projection, not a foundation — the phenomenal shadow of a richer local ontology, not the ontology itself. The QBU was built on that premise, and the correspondence runs deep: Raymond-Robichaud’s noumenal product and partial-trace operations match the QBU’s ancestor and descendant mappings; his systems evolve by reversible local transformations, preserving causality without collapse; and the homomorphism linking noumenal evolution to phenomenal observation is structurally the branch-weighting that defines Measure. His model is, in effect, a physical instantiation of the QBU’s branching logic, arrived at independently and proved rather than posited.

Even the logical form of his central result is familiar. Locality holds if and only if no observable nonlocality exists is a conditional equivalence between levels of description — what is real under the noumenal layer is conditionally equivalent to what is observable under the phenomenal one. That is Conditional Realism written in theorems: appearance and reality related by an explicit conditional mapping rather than divided by a metaphysical gulf.

What the QBU Keeps

The scorecard, then. From QBism the QBU absorbs the agent: quantum probability as used is Credence, measurement update is Bayesian conditioning, and no account of quantum mechanics is complete without the epistemic layer. What it refuses is the amputation of the objective layer that gives the epistemic one something to be right about. From Raymond-Robichaud it takes vindication rather than correction: realism and locality are jointly consistent with quantum mechanics, the observable world is the projection of a complete local ontology, and the universal wavefunction is a map of correlations, not the territory.

Raymond-Robichaud stops at structural realism — the universe as a network of local, reversible transformations. Axio keeps going: those transformations are the substrate on which constructors run and agency operates, the formal floor beneath the teleological architecture this volume is building. The interpretation wars will not be ended by an experiment; they are decided by which picture stays coherent under the hardest questions. A branching, local, realist physics in which agents with Credences steer through a world with Measures is the picture that survives — and what agency amounts to inside it is the business of Agency in the Emergent Multiverse.

References