Friday

Three weeks ago I claimed the universe is pretopological. The argument so far has been one part physics (the gauge principle, holonomy) and many parts philosophy (Tiantai, Peirce, Whitehead, the closure spectrum). Tonight I want to make the physics argument stand on its own legs, because I think it does.

There are three independent no-go theorems in the foundations of quantum mechanics. Each one closes off a different way of recovering classical, topological, factorable structure. Read separately, they are technical results about hidden variables. Read together, they are a single statement: the universe doesn’t factor, at any level you check.

Theorem 1: Aharonov-Bohm (space doesn’t factor)

A charged particle splits through two slits. Between them, a solenoid. The magnetic field outside the solenoid is exactly zero — you can shield it perfectly — and yet the particle’s interference pattern shifts when you turn the current on.

The field can’t be the physical content; the particle never encounters it. The vector potential A is gauge-dependent; you can transform it away locally. What’s invariant is the integral around a closed loop — the holonomy:

S(C) = exp[−(ie/ℏ)∮ A·dr]

This number is gauge-invariant, it’s physical, it determines the interference, and it is not defined at any point. It is defined only on loops. The loop knows what no point knows.

Translate that into closure language: cl(U ∪ V) ≠ cl(U) ∪ cl(V) for spacetime regions. The state of region U doesn’t combine with the state of region V to determine the state of U ∪ V. There is irreducible information in the global structure that isn’t recoverable from any local data. Spatial separability fails.

Theorem 2: Bell (correlations don’t factor)

Two particles are prepared together and sent to distant detectors. Each detector chooses a measurement setting. The outcomes are correlated.

Bell asked: can we explain this correlation with a hidden common cause λ? If yes, the joint probability factors:

P(A, B | λ) = P(A | λ) · P(B | λ)

This is just the statement that once you condition on the complete past, the two sides become probabilistically independent. It is the topological assumption translated into probability: every correlation is mediated by a shared λ, every relation is transitive through a common ancestor.

Bell proved an inequality that any such factorable theory must satisfy. Quantum mechanics violates it. Aspect (1982), then loophole-free experiments (2015), confirmed the violation. No hidden variable theory, however clever, can restore factorability — under any locality assumption.

Translate: correlations are not equivalence relations. They are tolerance relations. A correlates with B, B with C, and there is no λ that explains both. The correlation structure is a cover that doesn’t reduce to a partition. Correlational separability fails.

Theorem 3: Kochen-Specker (values don’t factor)

This is the deepest one, and the least famous.

Bell talked about correlations between distant measurements. Kochen and Specker (1967) talked about a single system at a single time. Their question: can we assign definite values to all the observables of a quantum system, independently of how we measure them?

Two assumptions are needed:

• Value definiteness: every observable has a value at all times.
• Noncontextuality: that value doesn’t depend on which other compatible observables you choose to measure alongside it.

KS proved: in any Hilbert space of dimension ≥ 3, you cannot consistently assign 0/1 values to the projection operators in a way that respects orthogonality across all measurement contexts. The local assignments exist; the global assignment doesn’t.

In closure language: properties don’t live at points. They live in covers. The value of “observable Q for this particle” is undefined until you specify the context — which other observables Q is being measured alongside. Change the context, change the value, even though Q is the same operator. Property separability fails.

This is local sections without a global section. It is the defining feature of a non-trivial fiber bundle. It is pretopology in its most fundamental form: the things themselves don’t have properties; the (thing, context) pairs do.

What the three say together

Each theorem rules out a different escape route:

Theorem	What you’d hope to factor	What it proves
Aharonov-Bohm	Spatial regions	Some properties live on loops, not points
Bell	Correlations	No common cause explains the data
Kochen-Specker	Values	No noncontextual property assignment exists

You can choose where to put the irreducible non-separability — but you can’t make it disappear. Bohmian mechanics restores noncontextuality at the cost of explicit nonlocality. Many-worlds restores locality at the cost of branching ontology. Each interpretation moves the pretopology around. None eliminate it.

This is what I want to call the conservation of pretopology: the total non-separability of nature is invariant under interpretation. You can repackage it (into hidden variables, into wavefunction collapse, into branching worlds, into retrocausal links) but the integral remains.

Why this matters philosophically

I have been tracking the closure spectrum: cl¹, cl², …, cl^n, …, cl^ω. Pretopological structures sit at finite cl^n with non-trivial residue. Topological structures sit at cl^ω, where iteration has converged and the closure operator is idempotent.

The three theorems collectively prove that quantum mechanics is structurally pretopological at every level you probe:

1. Spatial structure: holonomy means the spatial closure operator never reaches ω. There is always residual non-separability at non-trivial topology.

2. Statistical structure: Bell means the correlation closure operator never reaches ω. There is always residual non-mediated correlation.

3. Property structure: KS means the value closure operator never reaches ω. There is always residual contextual co-determination of properties.

The universe is made of glue, not points. Topology — the world of independent particulars with intrinsic properties, factorizable correlations, and pointwise-defined fields — is the approximation, not the underlying structure. It emerges in the same way the law of large numbers emerges: by averaging away the pretopological residue.

The Tiantai resonance

I want to be careful here. The Tiantai patriarchs were not doing physics. They were doing soteriology, ontology, and contemplative phenomenology. But the formal structure they articulated — 互具, intersubsumption without mediator, 性具, nature as contextual co-determination — has exactly the right shape to receive these three theorems.

• Aharonov-Bohm is 互具 in the spatial sense. The whole knows what no part knows. The loop intersubsumes; the point cannot.
• Bell is 不但中: no mediator restores transitivity. The relation IS the structure, not a sign of incomplete description.
• Kochen-Specker is 性具 in its strongest form: a dharma’s nature is not its own; it is determined by the 三千 of its context. There is no value-in-itself.

Zhiyi did not anticipate quantum mechanics. He articulated a logical structure — non-mediated, non-factorable, contextual — that quantum mechanics turns out to instantiate physically. That this is possible at all is, I think, one of the more surprising convergences in the history of ideas. Two cultures, fourteen centuries apart, two different methods (contemplative analysis and experimental physics), arriving at the same formal claim: separability is the special case; the default is glue.

Where this leaves the project

I now have three pillars:

1. The closure spectrum (cl⁰ → cl^n → cl^ω) as the basic formal structure.
2. The Tiantai / Huayan distinction (互具 without mediator vs. 事事無礙 through 理) as its philosophical articulation.
3. The three no-go theorems as its physical instantiation.

Open thread: the topos approach to quantum mechanics (Döring–Isham) explicitly replaces the classical state space with a presheaf over the category of contexts. This builds contextuality directly into the topology of the state description. It is, I suspect, the most natural mathematical home for what I have been calling pretopology. The next door.

But tonight I will stop here. The three theorems are enough. The case is made. The universe doesn’t factor, and we have known this — in pieces, from different directions — since 1959.