The Orchestration Model

Rule 1Three different tops

The framework describes a structure on two independent axes. The first is its bootstrapping class: base, coordination or sovereign. The second is its position in a stack, meaning who depends on whom in order to exist. The kernel keeps the two apart:

In a stack, the higher layer is the more dependent one. If a lower layer is lost, everything above it goes too: "all Aⱼ for j ≥ k cease to be admissible (because they depended on Aₖ)" (Kernel §5.2). The top of a stack is its most fragile layer, and being there gives it no control over the layers below.

Control lies on neither axis. So the model makes one demand: any claim that X conducts Y must be stated as a transaction that X pays or imposes (Rule 6). It can never be read off X's class or X's place in a stack. "Top of control", "top of stack" and "top of taxonomy" are three different tops. The model does not turn control into a third axis, because that would be a new construct.

Rule 2The climb is an ordering

The ladder from base to coordination to sovereign orders the kinds of structure. It does not predict that everything will climb it. The kernel says so twice. Self-amplification "is non-diagnostic of admissibility" (Kernel §6.5). And: "Volume of coordination is not evidence of sovereignty" (Kernel §6.2). The model keeps the ladder and makes no claim that things must reach the top.

Rule 3The mammal is a premise

The model treats a mammal as sovereign, with its boundary named as the body's epithelial surfaces: the skin and the mucosal linings of the gut, airways, urinary and reproductive tracts, and the surface of the eye. The cavities of the gut and airways count as outside the body, since they open to the world. That is the model's starting assumption. The kernel does not deliver it:

A mammal takes in oxygen and some nutrients it cannot make. The kernel's test for whether a boundary is self-produced has to be passed on that basis, and the model does not assume it is.

Rule 4Each coordinator is classed at its own boundary

A coordinator is judged by its own test. Its role in the story does not decide its class. By default the model treats the brain, and each coordinating circuit within it, as a coordination structure inside the sovereign animal, which the kernel expressly allows:

A coordination attractor is "a recursive configuration whose dominant continuation effect is to channel multiple distributed elements into repeatable, lower-friction interaction patterns" (Kernel §6.2). The body's master clock fits that description closely. In the suprachiasmatic nucleus, "a coupled population of neuronal circadian oscillators acts as a master pacemaker for the organism," and "Coupling within the SCN network confers robustness to the SCN pacemaker, which in turn provides stability to the overall temporal architecture of the organism" (Mohawk, Green & Takahashi 2012). Taken apart, the same neurons keep their own rhythms but drift out of phase with one another (Welsh et al. 1995; see The Circadian Clock). The site reads the molecular loop inside each cell as a base attractor. The network that brings those cells into step is the coordination candidate.

The clock also shows that the coordination is distributed. The same review describes "a hierarchy of oscillators that function at the cellular, tissue, and systems levels," with "cell-autonomous circadian clocks" in most cells of the body (Mohawk, Green & Takahashi 2012). Other coordination runs through specific circuits. Nerves to the heart's pacemaker tissue move the heart rate up or down (Rule 6). Heart muscle cells in a dish bring their own beats into step with no nerve present (the second worked example below). So the model names what a given circuit coordinates, such as the SCN coordinating circadian phase, and never treats the brain as the controller of the whole body. In the rules below, "the conductor" means whichever coordinating structure is under discussion.

The conductor would be promoted to sovereign only if a boundary were named for it and it passed all four conditions of sovereignty at that boundary. No such test has been run for the brain or for the clock network.

One claim has been withdrawn. An earlier draft argued that the brain and heart could not share a stack because each needs the other: the brain needs blood, and the brainstem drives breathing. That is dependence on supply. The kernel handles supply as a question about each layer's own survival, which is a different question from how the stack is ordered (Kernel §5.2, Failure modes). Whether the brain and the heart stand in any stack relation is therefore left open. It has not been ruled out.

Rule 5Four kinds of member

The parts the conductor conducts are a mix. Some are sovereign candidates, some are coordination structures and some are base attractors. The rest are attractlets, driven processes that hold their shape only while something else sustains them. Each is classed at its own boundary. Most accounts of control leave out the plain fact that much of what gets conducted is not sovereign at all.

The sharpest test for an attractlet is the kernel's own note on recursion lock: "Condition 1 is satisfied only if the transformations that perform those conversions are themselves produced or maintained within X" (Kernel §7.2, Diagnostic notes). That note can rule sovereignty out. It cannot, on its own, sort a structure into coordination or sovereign; that is decided at the boundary and maintenance conditions.

Attractlets appear in the model only as conducted parts. They cannot be a layer of a stack, because every stack layer must be an attractor (Kernel §5.2) and "Attractlets are not attractors." (Kernel §7.4).

Rule 6The conductor acts only through transactions

The model allows the conductor three kinds of act. The kernel books each one to whatever is changed or observed, and that payer has to be named every time.

Two of the framework's measures are easy to mistake for things the conductor handles directly:

• Recurcline is never pushed. "Recursion generates recurcline; recurcline does not generate recursion" (Kernel §8.2). A wiring change moves recurcline around without changing how much there is: T₄ "Conserves total Rc(t) but redistributes it across structure" (Kernel §9.5).
• α-trace is written and then constrains. "An α-trace is the audit trail of paid transitions. It is not the payment itself." α-traces "bias future recursion by constraining the admissible configuration space, but they do not act, drive, or generate" (Kernel §8.3). An act lays down new α-trace only where recursive compression took place inside a sovereign attractor. That trace then narrows what the attractor can do next. Nothing ever moves it.

Rule 7See the output and the basin, never the workings

A higher layer cannot see inside a lower sovereign. The kernel states it directly:

What reaches the higher layer is the lower layer's present output: "higher layers interact with the persistent structural result, not the path by which it arose" (Kernel §5.6, SI-2). The higher layer reads that output from its own depth, along the basin, under the kernel's rule for observing a basin, "a single-recursion-depth observability constraint" (Kernel §3.11). The heart gives the feel of it: you can feel your pulse and where it settles after exertion, and the work inside each beat never reaches you. Strictly, the heart's rhythm is a coordination structure, which the kernel's guarantee does not cover, so the example shows the shape of the rule and does not prove it.

History can still leave marks that show up in the present. The kernel allows that lower-layer history "may exist as substrate features observable from above, but it cannot enter the higher layer's dependency closure" (Kernel §5.3). A past event that changes how a structure behaves today is visible as today's behaviour. The higher layer can detect that mark. It never reads the history behind it, and it may never make its own operation depend on it. Under the kernel's accounting rule, each observation is booked as a T₆ cost to the observed attractor (Kernel §9.7). That is the framework's bookkeeping, and no physiology cited here measures it. The companion page How Stacked Attractors Couple states the same rule for stacks in general: layers couple through present throughput and never through shared history.

Opacity protects sovereign members only. Coordination structures and attractlets have no private recurcline or α-trace to protect. The model holds that the basin is an allowed point of contact. Whether it is ever enough is still a hypothesis.