Hex: Last time you gave me the six primitives as a concept — six dials on a control panel. Today I want the rule book. What does each definition formally say, and what does it formally exclude?
Lux: [warmly] Think of each definition as an ID card. It tells you exactly what the primitive is — and just as importantly, what it's not.
Hex: Then let's check IDs. P-one first — operator rewrite.
Lux: P-one is a rewrite of the substrate operator. You replace a Markov kernel P by a new kernel P-prime — or by a finite family of kernels. This changes the endomap, the thing that takes a distribution and evolves it forward one step. And it changes the induced empirical endomap that the Six Birds framework uses for packaging.
Hex: [nods] So P-one swaps the rules of motion.
Lux: Swaps the rules. And the definition is careful about one thing: this is a change in the operator itself, not an external schedule. Nobody outside the system is switching kernels according to a timetable. The rewrite happens because the system's own structure demands it. Think of it like a rule book that gets amended — not by an editor, but because the old rules produced a contradiction.
Hex: When does it demand it?
Lux: When macro dynamics fails to descend. If you try to define an evolution on the macro-level — on the quotient — and the result depends on which microstate you started from within a block, the macro-level evolution isn't well-defined. P-one says: fix the kernel until closure works.
Hex: P-two — gating.
Lux: P-two restricts the support graph. Formally: you set certain transition probabilities to zero and renormalize the remaining rows. Equivalently, you restrict the kernel to live on a subgraph.
Hex: Pruning pathways.
Lux: Pruning pathways. And this has a second-order effect that matters: by deleting edges, you change the cycle space of the graph. Fewer edges means fewer independent cycles. And since affinities live on cycles, P-two also changes what thermodynamic content the graph can carry.
Hex: So constraints don't just limit where you can go — they reshape the thermodynamic landscape.
Lux: Exactly.
Hex: [pauses] P-three. This is the one with all the caveats.
Lux: [carefully] P-three — autonomous protocol holonomy. Here's the formal setup. The state space is lifted: Z equals X times Phi. X is the original microstate space. Phi is a phase variable — an internal clock, not an external one. The phase evolves by its own kernel S. And conditioned on the current phase, the microstate updates by a phase-dependent kernel K-sub-phi.
Hex: So the protocol is part of the system. The phase variable isn't imposed from outside — it's an internal degree of freedom with its own dynamics.
Lux: That's the autonomous formulation. The definition explicitly says: no external schedule is assumed. Externally scheduled stroboscopic protocols — where someone outside the system toggles the rules on a timetable — are non-autonomous and fall outside the autonomy axiom A-A-U-T.
Hex: And the route mismatch that P-three measures?
Lux: The discrepancy between two routes: evolve then coarse-grain, versus coarse-grain then evolve. If those give different answers, there's route mismatch. P-three quantifies it.
Hex: But it's not a directionality certificate.
Lux: [firmly] The definition states this explicitly. P-three is a protocol-geometry diagnostic. By itself, it does not certify an arrow of time. Any arrow-of-time claim must be supported by an audit functional — which means P-six.
Hex: [writing] ID card for P-three: measures route mismatch, requires internal phase, no external clock, not an arrow.
Lux: That's the ID card.
Hex: P-four — sectors.
Lux: P-four is a conserved sector label. The support graph decomposes into disconnected components — blocks — and the kernel is block diagonal up to permutation. Evolution preserves the sector coordinate. Once you're in a sector, you stay there.
Hex: Protected variables. The things that don't change under the dynamics.
Lux: Protected variables. Think of electric charge or baryon number in physics — quantities that evolution respects absolutely. And when P-four holds, the theory index grows in a bounded way — linearly, not exponentially. That's what makes coherent refinement possible. Without bounded growth, the refinement chain could explode and the framework would lose its grip.
Hex: P-five — packaging.
Lux: P-five is an idempotent endomap e whose fixed points are the packaged objects of the theory. Apply e once: you get the object. Apply it again: nothing changes. The fixed-point set Fix-of-e is the collection of objects at that theory level.
Hex: [interested] And the paper says this is an endomap, not an order-closure. Why does that distinction matter?
Lux: Because an order-closure operates on a poset — it moves elements up to their closure. An idempotent endomap operates on distributions — it can act on probability measures, not just set elements. The framework needs this because packaging in physics takes a distribution of microstates and collapses it to a macro-distribution. That's an endomap operation, not a lattice operation.
Hex: So the math tracks the distinction between "closing a set" and "collapsing a distribution."
Lux: Precisely. And it matters when you compute defects — how far from idempotent the packaging actually is. The defect is defined on distributions.
Hex: [leans forward] P-six. The big one.
Lux: P-six is an accounting and audit structure. Formally: a certificate or functional that is monotone under coarse maps or packaging. Coarser views can't inflate the audit number.
Hex: One definition, but three instantiations?
Lux: Three canonical ones in the main paper. First: the information-feasibility order. This comes from the meta-theorem — limited access itself creates an ordering on what's observable. Second: path-space KL asymmetry with data processing. This is the arrow-of-time audit — the KL divergence between forward and reversed path measures, which data processing guarantees can't increase under coarse-graining. Third: the accounted graph 1-form with cycle integrals. This is the finite-graph version — log-ratio of forward to backward transition rates, integrated around cycles.
Hex: And P-six-drive is the third one?
Lux: P-six-drive is the specific ACC specialization: a non-exact log-ratio 1-form. Equivalently, nonzero cycle integrals — the system has genuine thermodynamic cycles that aren't just noise. This is the condition that makes "P-three needs P-six-drive" precise. Route mismatch is a geometric fact. Whether it carries thermodynamic weight depends on whether P-six-drive is active.
Hex: Has this actually been tested experimentally?
Lux: [nods] The particle paper runs exactly this experiment. They set up a substrate — a particle system — and toggle P-six on and off while holding P-three off. When P-six is off: audit proxies are near zero. When P-six is on: audit proxies activate — affinities become nonzero, motif diagnostics light up. Clean separation between null and drive.
Hex: [impressed] So the definition isn't just formal — it's experimentally separable.
Lux: In the emergence calculus framework's own experiments, yes. The definition is sharp enough that you can design a test: turn this switch, measure that number, check whether it separates.
Hex: [sits back] Six ID cards. P-one: swap the kernel, forced by descent failure. P-two: prune the graph, reshape feasibility. P-three: internal phase, route mismatch, diagnostic only. P-four: conserved sector, protected variables. P-five: idempotent endomap, objects as fixed points, not an order-closure. P-six: monotone audit with three instantiations, P-six-drive as the thermodynamic specialization.
Lux: [pleased] And every one of those exclusions matters. "Not an external schedule" for P-three. "Not an order-closure" for P-five. "Not a directionality certificate" for P-three. The rule book draws lines.
Hex: Sharp lines. I respect that. And notice — each exclusion is doing real work. Calling P-three "not an arrow" isn't hedging. It's a design decision that forces you to go find the arrow somewhere else — in P-six.
Lux: That's exactly the point. The definitions constrain each other. No single primitive overclaims.
Hex: Next time?
Lux: Episode twenty-nine — "Mapping to the Spine." We take these six definitions and connect them to the three-certificate backbone: stability, novelty, directionality. Which primitives serve which certificate, and where the wiring runs.
Hex: The wiring diagram behind the loop.
Lux: The wiring diagram.