Ffellonic Geometry and Natural Induction: Order without a Blueprint

Introduction

Two recent frameworks offer a suggestive point of comparison for how order emerges in complex systems. Ffellonic geometry, a hierarchical model of sphere packing in three-dimensional space, describes a 12-level progression from minimal connectivity toward a maximally coordinated lattice. Natural induction, developed by Richard Watson, Michael Levin, and Tim Lewens in "Evolution by natural induction" (Interface Focus, 2025), building on earlier work by Chris Buckley, Tim Lewens, Michael Levin, Beren Millidge, Alexander Tschantz, and Richard Watson in "Natural induction: spontaneous adaptive organisation without natural selection" (Entropy, 2024), explains how networked dynamical systems can spontaneously discover adaptive organization through relaxation under stress, without requiring natural selection.

One is geometric and physical; the other is biological and dynamical. This essay asks how far the resemblance between them actually goes — and argues that the honest answer is: real at the level of general mechanism, unestablished at the level of specific correspondence. The two are not yet shown to be the same process, but they may share a common ancestor worth naming directly.

The Core Mechanisms

Ffellonic Geometry: Geometric Relaxation

Ffellonic geometry begins with identical spheres attaching by touching. Each added sphere settles into the position that maximizes contacts while minimizing free energy, driving the system through twelve discrete levels — from a dyad (Level 1, one connection) through the icosahedron (Level 5, twelve-vertex coordination) to a maximally coordinated lattice (Level 12, twelve connections per sphere, matching the three-dimensional kissing number).

The progression is deterministic and dissipative: energy is irreversibly converted into stronger lattice bonds, exporting entropy to the surroundings as local order increases. Each level functions as a deeper energy minimum — an attractor the system settles into as new spheres arrive. As elsewhere in this series, one caveat belongs here: Level 12 is often described as a single, uniquely determined ground state, but FCC and the alternative close-packing HCP are extremely close in energy for hard spheres, and which is actually favored remains a live question in the colloidal-physics literature. Level 12 is best read as the class of maximal 12-fold coordination.

Natural Induction: Networked Relaxation

Natural induction occurs in networked systems where connections "give way slightly under stress" and the system experiences occasional perturbation. The network relaxes into lower-energy configurations in its state space, spontaneously forming adaptive organization. Crucially, no variation-selection-replication cycle is required at the global level for this to happen — adaptation can emerge from the system's intrinsic relaxation dynamics alone. Watson, Levin, and Lewens point to examples including phenotypic plasticity, symbiosis, and ecosystem restructuring, where organized order arises faster, or at scales, where Darwinian selection alone can't easily operate.

Unlike Ffellonics, this framework is explicitly general: it's stated to apply to any sufficiently plastic networked system, with heterogeneous nodes and interaction rules, not identical units in a fixed geometry. That difference matters for what follows.

A Shared Ancestor, Named Directly

Rather than asserting that these two frameworks describe the same process, it's more accurate — and more useful — to name the physical mechanism that plausibly underlies both: dissipative adaptation, the concept developed by Jeremy England, which describes how driven, far-from-equilibrium systems can be statistically biased toward configurations that absorb and dissipate more energy from their environment, without any selection process choosing them. Both Ffellonics' free-energy minimization and natural induction's stress-relaxation dynamics are consistent with this broader physical picture — and notably, England's work is cited in the natural-induction papers' own reference lists, suggesting the authors already see their framework as continuous with this line of physics.

This reframes the comparison usefully: Ffellonic geometry and natural induction aren't shown to be the same process, but they may be two applications — one narrow and geometric, one broad and dynamical — of a shared underlying physical tendency for driven, dissipative systems to settle into low-effective-energy configurations. That's a more modest claim than identity, and a more defensible one.

Where the Parallels Are Real, and Where They're Not Yet Established

Local rules and relaxation to attractors — this parallel holds at the level of shared mechanism, not shared specifics. Both systems have local rules governing how units adjust (attach, or give way under stress), and both settle into deeper minima over time as perturbations arrive. This is the strongest and most defensible point of contact between the two frameworks.

Order discovered, not imposed — also a fair parallel. In Ffellonic geometry, symmetrical structures emerge from the attachment rule rather than being designed in advance. In natural induction, adaptive organizations are discovered through relaxation rather than selected for by differential reproduction. Both are genuine instances of order arising from local dynamics rather than external design — this is a real structural similarity, independent of whether the two systems are otherwise equivalent.

Identical units — this one does not hold, and it's worth being direct about the contradiction rather than letting it stand unresolved. Ffellonics requires strictly identical spheres; natural induction is explicitly built to accommodate heterogeneous nodes with only similar (not identical) interaction rules. This is a genuine structural difference between the two frameworks, not a minor detail — it means natural induction is the more general case, and Ffellonics a much more constrained special case of a shared underlying idea, if the idea is shared at all.

Multi-scale hierarchy — the parallel here is looser than it first appears. Ffellonics' hierarchy is a fixed, deterministic sequence of twelve geometric levels. Natural induction's multi-scale operation (cells to tissues to organisms) is open-ended and doesn't follow a predetermined sequence of discrete stages the way Ffellonics does. Both are "hierarchical" in a broad sense, but the word is covering two different kinds of structure — one a fixed ladder, one an open-ended nesting.

Induction preceding selection — this comparison doesn't currently map cleanly and is worth flagging rather than repeating uncritically. In Watson, Levin, and Lewens's account, induction discovers adaptive organization that natural selection may later genetically canalize — a specific two-stage biological process. Ffellonic geometry has no clear analogue to the second stage: there's no process that comes after a packing forms and "fixes" it the way selection fixes a genetically favorable trait. Physical law isn't a fixation mechanism that acts on an already-formed structure; it's what produces the structure in the first place. Until a real analogue to the canalization stage is identified in the geometric case, this parallel should be set aside rather than asserted.

What Would Be Needed to Say More

The honest state of this comparison is: a plausible shared physical ancestor (dissipative adaptation), two genuinely overlapping structural features (local relaxation to attractors; order discovered rather than imposed), and at least two places where the parallel breaks down or hasn't been worked out (identical vs. heterogeneous units; the missing canalization analogue). Turning "these frameworks may share a mechanism" into "these frameworks describe the same process" would require something the current comparison doesn't have: a demonstration that natural induction's dynamics, applied to a system of literally identical units in 3D Euclidean space, actually converges on something resembling the Ffellonic hierarchy specifically — rather than simply being consistent with the same broad class of dissipative, energy-minimizing behavior that a great many physical and biological systems exhibit.

Conclusion

Ffellonic geometry and natural induction are not yet shown to be two descriptions of the same fundamental process — that claim outruns what either framework, on its own terms, establishes. What's more defensible, and still genuinely interesting, is that both are plausible instances of a broader physical tendency — dissipative adaptation — toward low-effective-energy configurations under stress or perturbation, one applied narrowly to identical spheres in Euclidean space, the other applied broadly to heterogeneous biological and ecological networks. Whether the narrow geometric case is best understood as a special instance of the broader dynamical one, as this essay suggests, or as a separate phenomenon that merely resembles it, is an open question — and one worth investigating directly, by asking whether natural induction's own relaxation dynamics, run on sufficiently uniform units, actually reproduce anything like the Ffellonic hierarchy, rather than assuming in advance that they must.