Canonical · Relation to Adjacent Work

Relation to Adjacent Work

Where this framework agrees with its neighbors, and where it parts company

Principia Attractum extends nonlinear dynamics; it does not replace it. Much adjacent work describes the same phenomena — attractors, basins, stability, self-organization — correctly and rigorously. This page states, work by work, exactly where the framework agrees and where it draws its own line. The pattern varies by neighbor. Some — the dynamical-systems lineage — describe structures that settle, and this framework adds the layer that asks what it costs to keep holding together and who is paying. One — autopoiesis — is not a neighbor the framework extends but a predecessor it stands on: the ground beneath its central condition rather than a case it reaches past. This page states each relationship on its own terms, including the debts.

Autopoiesis

Maturana, H. R., & Varela, F. J. (1972/1980). Autopoiesis and Cognition: The Realization of the Living. D. Reidel. [verify edition/pagination against primary source before publication]

This is the closest neighbor the framework has, and honesty requires putting it first. An autopoietic system, in Maturana and Varela's sense, is one that continuously produces the very network of processes that produces it — and, crucially, produces and maintains the boundary that distinguishes it from its medium. A living cell is their paradigm case: it makes the components that make the membrane that contains the components. Self-production, closure, and a self-generated boundary are the heart of the idea. If any adjacent work can be said to be underneath this framework's central condition, it is this one.

Where the framework agrees — and owes a debt. The framework's Condition 3, that a sovereign attractor must produce and defend its own identity-bearing boundary, is standing on ground Maturana and Varela cleared. Their self-produced boundary is this framework's boundary retention; their operational closure is close kin to recursion lock; their organism-medium distinction is the same inside/outside relation the framework's distinction-persistence primitive presupposes. On the core claim — that the living thing is the one that makes and holds its own boundary, and is not merely held together from outside — the two are in near-complete agreement. The engine-versus-cell intuition this framework leans on is the autopoietic intuition.

Where the framework adds its own layer. Three differences, none a disagreement so much as an extension:

1. Cost is made explicit. Autopoiesis emphasizes self-production and closure; this framework insists, through Condition 4 and the throughput constraint, that self-production is never free — it runs at ongoing cost that must be paid from outside the boundary, always. Autopoiesis is compatible with this, but the framework foregrounds the bill and asks who settles it.

2. A binary threshold with named failure. This framework makes sovereignty an all-or-nothing four-condition test with an explicit account of irreversible loss. Autopoiesis is more often stated as a qualitative characterization of the living; the framework hardens it into a pass/fail classification with a defined boundary-violation event.

3. Sequence, not time; and no prestated space. The framework carries its own substrate commitments — order rather than time, and a space of possibilities that is not fixed in advance — that autopoiesis does not itself require. These are additions, not corrections.

Autopoiesis says the living thing makes and holds its own boundary. This framework agrees, then adds: and it pays, continuously, to do so — and the day it can no longer pay, the boundary is lost and is not inherited back. The idea behind Condition 3 — the self-produced, self-defended boundary — is Maturana and Varela's intellectual lineage; Condition 4 and the loss account are what this framework carries in on top.

The relationship is best stated plainly, because a reader steeped in the literature will test for it: this framework does not claim to have originated the idea of the self-produced boundary — that lineage is Maturana and Varela's, and this is an acknowledgment of intellectual debt, not a transfer of authorship (the framework's own canonical text is authored and governed separately). It claims only to have placed that idea inside a wider persistence accounting — cost, throughput, irreversible loss, and a measurement layer that exists only while the boundary is actively held. Where autopoiesis characterizes the living, this framework tries to give the conditions and the bill.

Random Boolean Networks

Gershenson, C. (2004). Introduction to Random Boolean Networks. In Bedau et al. (eds.), Workshop and Tutorial Proceedings, Ninth Int. Conf. on the Simulation and Synthesis of Living Systems (ALife IX), pp. 160–173. arXiv:nlin/0408006v1. [verify arXiv identifier, venue, and pagination against primary source before publication]

Random Boolean networks (RBNs), the N–K or Kauffman networks, are a natural neighbor. A network of N nodes, each taking a value from K inputs through a fixed (“quenched”) logic function, is run forward deterministically. Because the state space is finite, the network eventually repeats a state and has, in Gershenson's words, “reached an attractor”; the set of states flowing into it is “the attractor basin.” The paper is a careful tutorial, and its care matters here: it says living organisms could be constructed from such elements and that certain networks have analogies to life — it does not claim an RBN attractor is a living thing. That restraint is why the framework agrees with almost everything it actually asserts.

Where the framework agrees. The attractor-and-basin machinery is exactly the dynamical vocabulary this framework borrows: a deterministic map, a finite state space, states flowing toward a cycle they return to. The K bounded inputs are adjacency made literal — influence restricted to neighbor relations. The quenched transition function is a field of asymmetric admissibility — some transitions permitted, their reverses not. And the paper's perturbation analysis, where in the ordered regime “a perturbed network returns to the same path of the normal network,” is return-to-identity after a bounded shock: the shape of the return, the same effect you can watch in the live Eco 1.0 simulation, a small in-browser ecosystem elsewhere on this site that recovers from an imposed shock. On the dynamics, the two are in agreement.

Where the framework parts company. Not with anything Gershenson claims — with the layer his tutorial stops beneath. Three things an RBN does not have:

1. The persistence question does not apply. An RBN attractor is a cycle in a state graph — a mathematical property of a closed, deterministic map. It is worth being exact about what it is not. It is not a sovereign attractor: nothing is paid to occupy it. But it is also not an attractlet — an attractlet is stability supplied by an external agent, and an RBN attractor has no external agent propping it up; the map is closed, and nobody is holding it. It is a third thing. This framework's central question — who is paying to keep it there? — is a category question, and for a cycle in a state graph the honest answer is that the question does not apply. That is not a way of sorting the RBN attractor into an existing bin; it is the precise sense in which the model does not reach the layer this framework works in. An RBN attractor makes no physical persistence claim at all, and so is neither sovereign nor attractlet — it is a bare mathematical fixed point (or cycle), and the who-pays question is the boundary between its layer and this one.

2. A pre-stated space. The nodes, the wiring, and the logic tables are all fixed in advance; the state space is enumerable before the network runs. This is the pre-stated landscape the framework declines to assume for open-ended living systems — the same tension named on the basins page between a fixed terrain and one carved by the running of the system. Gershenson himself notes the edge of it: “nature does not exhaust all possible configurations.”

3. Externally forced regularity is exogenous by construction. This is a narrow point about one class of result, not a claim about RBN attractors in general. When the paper's control methods force a chaotic network into a pattern — using “a periodic function to drive” it, with “a high percentage of nodes” controlled — the regularity that results is imposed: the network does not hold that pattern itself, it is held there. The paper is candid about exactly this. So where an RBN's order is externally driven, it is the exogenous case, and would read as an attractlet if it were a physical system. This says nothing about undriven RBN attractors, which fall under point 1; it addresses only the forced case, and only to note that forced order is borrowed order.

An RBN attractor has a basin but does not pay a bill. Ask this framework's question — who is paying to keep it there? — and for an RBN the answer is “nobody, it runs for free.” That free-ness is the whole distance between a Boolean-network attractor and a sovereign one.

One caution belongs to readers, not to Gershenson: the well-known result that RBN attractor counts roughly track cell-type counts is a useful analogy, and the paper lists its many caveats plainly (revised genome size, scale-free rather than homogeneous topology, biased functions, and the decisive point that “genes do not march in step”). To slide from “an RBN attractor models a cell type” to “an RBN attractor is a self-maintaining cell” would be a category error — a cell type is a sovereign attractor that pays its own maintenance and makes its own boundary, while an RBN attractor is a free cycle of a fixed map. The tutorial stays on the safe side of that line; this framework simply names the line.

A proposed extension · offered, not asserted What an RBN would have to be given to make the who-pays question apply. The section above draws a line and stops. But the line suggests its own experiment, and it is one that runs entirely inside Gershenson's own apparatus — the N–K network, the quenched logic functions, the finite state graph, the perturbation analysis — with one layer added on top. The claim of the section is that an RBN attractor runs for free: nobody pays to keep it there. So the natural next experiment is to make it pay, and watch what changes. This is a research direction, not a framework result; it introduces nothing to the canon and predicts nothing in advance. It is a question posed to the model.

The question

Take a standard RBN in the ordered or critical regime, where a perturbed network “returns to the same path of the normal network.” That return-to-identity is already visible in the model. Now attach a running cost to occupying an attractor cycle, a finite supply that pays that cost, and a rule that ends the cycle if the supply runs out. Does the free cycle of a fixed map begin to behave like something that has to hold itself together — and at what point, if any, does the who-pays question stop being inapplicable and start having an answer?

A possible simulation

Start unchanged. A standard RBN: N nodes, K inputs each, a fixed random logic table per node, run deterministically to an attractor cycle with its basin. Nothing new yet — this is the tutorial's own object.
Add a maintenance ledger (the bill). Give the network a scalar reserve S. Each update step debits a maintenance cost c from S. This is the framework's Condition 4 made literal: occupying the cycle now costs something per step.
Add a throughput supply (who pays). Credit S from an ambient inflow each step — the model's stand-in for the throughput condition, that nothing persists closed. Run two arms deliberately: an exogenous arm where the inflow is supplied by the experimenter on a fixed schedule, and an endogenous arm where the inflow is gated by the network's own state — e.g. only certain attractor cycles “earn” inflow, so the network's own dynamics determine whether it can pay its own bill.
Add a loss event (the boundary that can be violated). If S falls to zero, the network stops — it does not simply wander to another cycle; the run terminates, and any restart must re-initialise rather than resume. This makes loss irreversible in the run, the model's echo of the framework's account of boundary violation and non-inherited identity.
Reuse Gershenson's perturbation analysis, but ask a costed question. Flip b nodes as he does, but now measure not only whether the network returns to its path, but whether it returns before the reserve is exhausted — recovery becomes a race between the return-to-identity he already measures and the depletion of S. This turns his ordered/critical/chaotic distinction into a costed survival question.

What each arm is expected to isolate

The exogenous arm should read as the attractlet case: the pattern is held there by a supply the experimenter controls, and cutting the schedule collapses it. This is the costed version of the paper's own candid point about externally forced regularity — forced order is borrowed order.
The endogenous arm is the interesting one: if some attractor cycles gate their own inflow well enough to keep paying while others starve, the model now sorts its cycles by who pays to stay there rather than by cycle length or basin size alone. That sorting is the who-pays question becoming answerable inside the model — the point at which a bare mathematical fixed point acquires the beginnings of the persistence layer the framework works in.

What this would and would not show. It would not show that an RBN cycle “is alive” or is a sovereign attractor — the boundary in a costed RBN is still supplied by construction (the wiring and logic tables are fixed in advance, per point 2 of the parting above), so Condition 3, the self-produced boundary, is exactly what this design still does not reach. What it would show is a clean, minimal demonstration of the line the framework draws: the same network, run free versus run costed, and the difference the bill makes to which cycles survive. It would give Gershenson's model a way to be asked the framework's one question — who is paying to keep it there? — and to answer it, arm by arm, from its own dynamics. The remaining, harder direction it points toward, letting the wiring itself be produced and repaired by the running network so the boundary stops being pre-stated, is where the model would begin to approach Condition 3, and is noted here only as the direction, not the experiment.

This harness already exists. The four pieces above — a maintenance ledger, a throughput supply, an irreversible loss event, and the exogenous-versus-endogenous split — are not a new apparatus to be built. They are the design of this site's own Eco 2.0 harness, where Eco 1.0 is an attractlet that pays its own metabolic bill each step (Condition 4, met) but is handed its boundary (Condition 3, the gap). The costed-RBN proposal is that same harness ported from an ecology onto Boolean dynamics — which is the honest way to read it: not a fresh build but a port of a rig already running, and it inherits the rig's exact ceiling. Eco 2.0 hits Condition 3 by trying to make the boundary self-produced; the RBN version, with its wiring fixed in advance, does not attempt that step, and so stops one condition short in the same place, for the same reason.

The map, and what is still to come. The sidebar is the full inventory of neighbors, and every entry in it resolves to a treatment — two on this page (autopoiesis and random Boolean networks, given full treatments above) and the rest on their own pages: prions, autocatalytic sets, the mitochondrion, Waddington's landscape (with Kauffman's generated-possibility critique treated alongside it), biomolecular condensates, LLMs and agents, Hopfield networks, and bifurcation theory. Each is read the same way: what the framework shares with the neighbor, and the one place it draws its own line.

Three neighbors are named but not yet written, and are listed as In preparation in the sidebar. They are the honest gaps:

Viability theory (Aubin) — survival under constraint; close kin to admissibility, and the other neighbor a serious reader will expect to see. Highest priority to add next.
Control theory — Lyapunov stability and feedback; shares the stability-margin instinct but is silent on who supplies the control.
The demotion of fundamental time (Barbour; causal-set theory) — the neighbor for sequence, not time.

Adjacent-work assessments state where Principia Attractum agrees with and departs from neighboring frameworks. They introduce no constructs and modify no canon; they locate the framework relative to its field, and — where a neighbor's apparatus invites it — may offer a proposed extension of that neighbor's own work, marked as offered rather than asserted and making no canonical claim. Quotations are taken from the cited primary sources. Where a source is paraphrased rather than quoted, the paraphrase is marked as such.

← Back to Principia Attractum, HTML Edition