Hex: The myth — "build a good enough macro description and it captures everything about the micro level." Lux, true or false?
Lux: [leaning forward] False. By an exponential margin. The finite forcing lemma says almost nothing about the micro level is expressible from the macro. Today we bust this with pure counting.
Hex: Set the stage. What's a predicate?
Lux: A predicate is the simplest question you can ask about a microstate. Each microstate gets a yes or a no. Technically: a function h from the microstate space Z to zero or one. There are N microstates, so there are two to the N possible predicates — two to the N different yes-or-no labelings of the micro world.
Hex: And some of those questions can be answered from the macro level?
Lux: [nodding] The ones that respect the coarse-graining. You have a theory — a lens — that maps microstates to macro states. N microstates, K macro states. A predicate is definable from the theory if every microstate in the same macro block gets the same answer. If you can't tell two microstates apart at the macro level, a definable predicate can't tell them apart either.
Hex: Give me a picture.
Lux: [gesturing] Think of a dictionary with K entries. Each entry covers a whole cluster of microstates. A definable predicate is a question the dictionary can answer — it only needs to look up which entry you fall under. A non-definable predicate is a question that requires finer resolution than the dictionary provides. It needs to distinguish microstates within the same entry.
Hex: How many definable predicates are there?
Lux: [counting on fingers] One binary choice per macro state. K macro states. So two to the K. That's the counting lemma, and it's tight — no approximation, no hand-waving. Exactly two to the K. The proof is one line: a definable predicate assigns one bit to each block, and the blocks are determined by the lens.
Hex: And the total?
Lux: Two to the N. The ratio of definable to total predicates is two to the minus N minus K. Let me make that concrete. Say you have a hundred microstates and ten macro states. Two to the ten is about a thousand definable predicates. Two to the hundred is about ten to the thirty — a one followed by thirty zeros. The fraction you can express from the macro? Roughly ten to the minus twenty-seven.
Hex: [stunned] That small.
Lux: [firmly] Smaller than the ratio of a single proton to the observable universe. And this is the theorem — the finite forcing lemma. The probability that a randomly chosen predicate is not definable from your theory is one minus two to the minus N minus K. For any reasonable N and K, that's essentially one. Non-definability is the generic case.
Hex: The paper calls this "generic." That's a loaded word in mathematics.
Lux: [nodding] Deliberately loaded. Borrowed from mathematical logic. In Cohen [KO-en] forcing, you build generic extensions of a mathematical universe — new objects that can't be defined in the old language. The finite forcing lemma is the combinatorial analogue. A random predicate can't be defined from the old theory. But the paper is careful — it explicitly says this is a combinatorial proxy. It's not claiming set-theoretic independence. The structural parallel is exact; the logical depth is different.
Hex: [pressing] So "generic" means "overwhelmingly probable" in this context?
Lux: [precisely] In the finite setting, yes. Generic means: the complement has an exponentially small probability. Generic extensions are the norm. Definable extensions are the needle in an exponentially large haystack.
Hex: [leaning forward] Now here's where it gets interesting. What happens when you adjoin one of these non-definable predicates to the theory?
Lux: [with emphasis] You get a strict refinement. The refined lens — the original theory plus the new predicate — distinguishes at least one pair of microstates that were previously indistinguishable. The corollary is explicit: with probability one minus two to the minus N minus K, the extension is strict. It genuinely expands what the theory can see.
Hex: And there's an even stronger result?
Lux: The "nothing stays constant" lemma. Not only is h non-definable — it splits every macro block. The probability that h is constant on a block of size m is two to the one minus m. For blocks of size ten and ten macro states, the probability that h splits every single block exceeds ninety-nine point nine eight percent.
Hex: [leaning back] So the new predicate doesn't just break one distinction — it breaks all of them.
Lux: [spreading hands] Almost surely. Every block gets cracked open. The new predicate sees differences the old theory was blind to — everywhere, not just in one corner of the state space.
Hex: Why does this matter for the emergence calculus?
Lux: [shifting posture] Because it's the anti-saturation mechanism. Remember closure — P5, the packaging primitive? Closure saturates. Apply it twice, you get the same result. Idempotence. The system reaches a fixed point and stops. But extension — adjoining a new predicate — breaks saturation. It forces the theory to grow. That's how the framework models ladder-climbing: not by infinite internal iteration, but by strict extension from outside.
Hex: So closure builds a ceiling, and extension punches through it.
Lux: [pointing] Exactly. And the finite forcing lemma tells you: the ceiling is thin. Almost any extension punches through. The hard part isn't finding new information — it's the extension itself, the act of adjoining a new predicate and rebuilding the theory around it.
Hex: [shifting topic] The quantum theory paper picks this up?
Lux: [animated] Beautifully. The QT paper maps definability directly to quantum contexts. When you change measurement basis — say from the z-basis to the x-basis — you're not revealing a pre-existing value in the same record language. You're extending the record algebra itself. A strict extension. The new basis defines a predicate that was not definable from the old basis.
Hex: And that's how the emergence calculus reads contextuality?
Lux: [with emphasis] Precisely. Two incompatible packaging maps — dephasing in two different bases — don't commute. Alternate them and states drift toward the maximally mixed state, the intersection of their fixed points. The noncommutation is the structural signature. The definability rarity is the reason it's unavoidable — there are exponentially more non-definable predicates than definable ones.
Hex: [thoughtful] And the Lean anchor?
Lux: The counting lemma is machine-verified. The Lean statement — card DefPred equals two to the card X under a surjective lens — lives in DefinabilityCount dot lean in the companion code. There's also a generalization for non-surjective lenses and a specialization for product projections. The Six Birds project's pattern holds: the algebra is machine-checked, the numerics are reproducible, and the gap between the two is clearly flagged.
Hex: [nodding slowly] So the viability work from the agency paper connects here too?
Lux: [drawing a connection] Yes. The agency paper's viability iteration computes a greatest fixed point — a stable safe set. But what's expressible at that fixed point depends on the layer's predicates. Definability rarity means the expressive power of any fixed layer is inherently limited. You can stabilize within a theory, but you can't express everything from within it.
Hex: [sitting back] Myth busted. Macro descriptions are exponentially incomplete. Almost any new predicate breaks the old theory. And the framework treats this as a feature, not a bug.
Lux: The emergence calculus embraces it. Theories are ladders, not ceilings. Extension is generic. Saturation is local. And the exponential gap isn't a flaw in the theory — it's a structural fact about what coarse-graining can and can't do.
Hex: Next time?
Lux: Episode forty-eight — "Balanced-Atom Route: Definitions and the Kernel-Mass Hinge." From what theories can't say to what atoms can hold.
Hex: From the dictionary to the vault.
Lux: From words to real weight.