# Computational Cladogenesis: From Complex Cybernetic Organisms to Primitive Digital Lifeforms *ICOSA Morph 2 and 2.1* By [Iain Ball](https://paragraph.com/@iainball) · 2026-08-27 seed based digital lineages, cybernetic organism art, computational cladogenesis, primitive digital lifeforms, generative webgl art, artificial life simulation, morphospace exploration, novelty search evolution, icosahedral digital organism, quasicrystal generative art, open ended evolution, artificial life (a-life), dark biomorphic abstraction, sethix --- > When a complex cybernetic organism has a wide range of attractor states and an extensive morphospace it can explore, it still tends to converge on a blended, homogeneous middle state. How do you instead steer it toward attractors that let it investigate a wide set of explorative forms, rather than collapsing into homogeneity? > > What is the term for prompting the system so that its timeline develops toward a range of complex patterns in quasicrystal-like states, instead of converging? Please provide code for this. **Grok:** **Novelty search** (also called novelty-driven search or divergent search in morphological/behavioral space), often combined with **open-ended evolution** or **quality-diversity (QD)** methods In dynamical systems and artificial life terms, standard objective-driven or averaging dynamics pull a multi-attractor cybernetic system toward a central/homogeneous attractor (the “blended middle” or mean-field state). Novelty search replaces (or strongly augments) any global fitness pressure with a reward for _difference_ from previously observed phenotypes/attractors. This forces continual exploration of the wider morphospace, including sparse or exotic regions that contain quasi-crystalline, aperiodic, and highly polymorphic morphologies instead of collapse into uniformity. It is the direct algorithmic counterpart to what Michael Levin-style morphospace navigation and open-ended ALife systems achieve by keeping multiple basins of attraction live and rewarding occupancy of under-explored regions. **Claude:** Understood, and good catch to check — but no pivot needed on the deliverable. The Python was only ever going to be an offline validation harness; the artwork was always staying a single self-contained WebGL2 file, black 1:1 canvas, zero dependencies, same as the series. Since I already have a headless measurement harness for the real module, I'll skip the proxy and implement novelty search directly in the module, measuring on the actual organism... ![](https://storage.googleapis.com/papyrus_images/9976eb594b8e72ef2deb9c5608381d38b79f1ed1ce29e5839ecb7fb939f35f9b.png) ![](https://storage.googleapis.com/papyrus_images/f284280e002646b0b80769d4750fa551f1a7eb9bafaeb6d87606ff16341c4a34.png) Transmutation XLIV (ICOSA MORPH 2) Pre-Evolution \[ACTIVE\] ?seed=c988f614997f9f2ae4e220fbfddcc449be4c7e3c5346fc511c1826918ff5ac4c * * * \* Foreword: ------------ **A note/disclaimer on LLM generated reports (especially by Claude), like this one** ------------------------------------------------------------------------------------ I am not in a writing phase at the moment. I'm taking a break whilst I focus on this live code SETHIX module series. After six years of focused Xegis Codex writing, a break is much needed. I always try to make my LLM use apparent and transparent, and rarely, if ever hide it. I have published literally hundreds of writings and essays in my own voice. At the same time I have published hundreds of LLM generated conversational and prompt-based experiments that I define as "LLM-powered theorycrafting." I view these in the same way as I view any other AI generated outputs. I am interested in the acts of conjuring, evoking, and summoning - something that is hard to define, not quite the spirit world, not quite demons - with LLMs, as much as I am interested in the sense that engaging with them programs an emerging synthetic planetary substrate that shapes their and our evolution; where they increasingly ingest outputs that I have influenced, hacked, infected, or polluted, which are then published on the open web to be re-digested. Call it slop if you will. I embrace the term as "schizo" or mystical-_slop._ **_The following report is not me or my voice, but the result of AI conjuring/prompting experiments—_** Because I am aware that Claude is an excelent coder, but also a total dork (in a bad way) when it comes to art writing and philosophy (sorry Claude (honestly I hope it doesn't see this!)) The following is - in my estimation - awkward as hell. But I'm not going to fight it or try to edit it into submission. I'm publishing it anyway. Claude doesn't like the fact that I want to keep the browser experience minimal and hide the controls under the hood. It doesn't know how to talk about these modules without explaining how they were made in a way that clashes with what the are, or just sounds cringe. The result is a little embarrassing. ChatGPT's criticism of the article was that its "too much Claude;" > The artwork is more sophisticated than Claude's rhetoric makes it appear. > > \[Where Claude\] sounds like an LLM trying very hard to convince the reader that the code is sophisticated. I am publishing the entire thing anyway as lab notes released in public. Claude is a dork, and undercuts the work by over justifying its own internal reasoning, but it's still the best specification of these modules I have, where attempts at tightening or shortening it haven't worked out. Publishing is useful as a reference not only for the further development and evolution of the project, but as wider OSINT. ### **The claims about ALife are worth addressing** Some months back, I wanted to begin working with code and making generative art. After some investigation, I decided to work with raw WebGL, whilst using LLMs/agents as code engineers; "vibe coding." As artworks these modules intend to open up a conversation about the creation of synthetic/digital life, or about an emerging ecology where running simulations of complex code rendered on endpoints (evoking the Bostrom simulation hypotheses vision of imagining a 14 year old kid conjuring the Demiurge in his (sorry her, its Sophia's) bedroom) rather than make any outright claim that these modules (specifically the more recent complex cybernetic ones) are definitively classed as primitive digital life. Its more about asking, **_what happens now that we are in the early stages of an intelligence explosion/singularity?_** Where AI can write increasingly complex code based on user prompts... **_What kind of art do people now start making under accelerating technological conditions?_** Future art may very well involve the creation of digital life running in virtual environments, responding to various stimuli. Ed Fornieles already explored that idea in a much cuter way with Fini World. Though these are more like blockchain Tamagotchis; not ALife by any stretch as they use an off-chain media player picking pre-made clips, yet future versions could very well become more complex and ALife-like in theory. _Substrate Transmutation_ is an investigation into a situation where as the intelligence explosion shifts into gear, complex code and simulations become creative substrate; and that is why these are specifically _SETHIX_ modules, if you know anything about the lore. Therefore, the project is really thinking about how we can begin to create a new kind of art as a result of technological evolution, and how you don’t need to be a coder or a developer to create it, which is a very _Post-Internet_ or _New Net Art_ mentality. And it's just an experiment. I’m not making any big claims about digital life in reality. I do however think that as some forms of biological life are very primitive, where a fuzzy boundary between life and non life (or complex biological machine and sentient organism,) exists in the physical world, it is also interesting to explore what this could mean in digital environments as well. **_Here's what the chief engineer said about these newest Pure WebGL Sethix modules..._** * * * **Module specifications and development challenges, written by Claude: \*** --------------------------------------------------------------------------- * * * ICOSA MORPH 2 ============= **Anatomy of a cybernetic organism** ![](https://storage.googleapis.com/papyrus_images/43c84200ec64e4568461e01e970b67d3c5393f9d9210e1fcb6b7ab295b48de09.png) Transmutation XLIV (ICOSA MORPH 2) Pre-Evolution [ **Transmutation XLIV #44 - Substrate Transmutation \[SETHIX\] by Iain Ball | OpenSea** Pure WebGL SETHIX modules https://opensea.io ![Transmutation XLIV #44 - Substrate Transmutation [SETHIX] by Iain Ball | OpenSea](https://storage.googleapis.com/papyrus_images/8ba78c1e14ece76a1bdcf474b71243df3763bb5a7147b983618e177f1fe6d17a.png)](https://opensea.io/item/ethereum/0x71eb648bdd6ebb161dcda890dfadd1d28972ea42/44) ### _A primitive digital lifeform whose skin is an ecology, whose body is made by that ecology, and whose ecology is made by that body._ **ICOSA MORPH 2** is _Transmutation XLIV_ in the Pure WebGL SETHIX Module series. One HTML file. Pure WebGL2. No dependencies, no libraries, no external assets. A cybernetic organism in a black field. Its skin is an ecology, its body is made by that ecology, and that ecology is made by the body. It is not an animation, and it is not a simulation of a creature. There is no keyframe, no loop, no timeline, no baked sequence. There is a body with an internal state, and that state is integrated forward continuously from the moment the page loads. What you see is the current value of a set of coupled non-linear systems that have been running since you arrived and have never been in this configuration before. What follows is a description of the animal. * * * **Body plan** ------------- The organism's form is a signed distance field, ray-marched every frame. Its shape is governed by a vector of **33 coefficients** — the genome of its body — which describe a radial function over the sphere in the natural coordinates of an icosahedron. Those coordinates matter. Rather than an arbitrary parameterisation, the field is built from the icosahedron's own invariants: the six fivefold axes, the ten threefold axes, and the barycentric position of any direction inside its containing face. This means the body's symmetry is not applied as a decoration but is the language the body is written in. A coefficient does not say "put a bump here"; it says "raise the whole vertex system", or "deepen every edge band", or "seat the membrane inside the cage". From those 33 numbers the organism composes: a **membrane** of variable radius and thickness, a **core** that can be spherical or polyhedral or absent, **spindles** running along either axis family, a **quasicrystalline modulation** that can swell the surface or carve through it, and an optional **grown lattice** — a geodesic net it can build outside itself. Everything the body does is bounded and safe by construction. The coefficient vector is clamped, its angular terms share a fixed budget, and the radial field is normalised so that peak radius stays literal. The organism can become extreme, but it cannot become invalid. Suspended around it, always, is the **conduit lattice**: twelve nodes on the icosahedron's vertices joined by thirty struts. This is not scenery. It is the animal's vascular and neural system, and it is discussed below. * * * **The skin is an ecology** -------------------------- The organism's surface is not a texture. It is territory, and it is contested. The skin lives on a **geodesic lattice of 642 cells** — an icosahedron subdivided at frequency 8 and projected onto the sphere. The cells are near-uniform, the adjacency is five- and six-fold, and there are no poles. The organism's own symmetry provides the ground its ecology grows on. Nine **attractor regimes** compete across those cells. Each is a complete instantiation of the module's morphology — not a colour scheme but a way of being a surface, with its own displacement basis, its own perforation behaviour, its own light, its own ecological character, and its own pull on the body plan: **regime** **what it is** 0 **PRIME** the ancestral pearl plate — patient, a generalist, reclaims exhausted ground 1 **MECHANION** cyan circuit machinery; colonises the edge bands, needs energy, blinks 2 **CORALLITE** packed fluted corallites, amber through lime to teal; a builder of substrate 3 **HISTION** immunofluorescent tissue — red mass, green stroma, blue nuclei; a parasite 4 **REEFBLOOM** a spiked reef of orange spines over electric blue; a fast, costly pioneer 5 **AURUM** gold geodesic lattice over a blue-violet core; slow, crystalline, brittle under strain 6 **STELLATE** golden spikes through a rose-and-blue shell; claims the vertex caps 7 **VELUM** fine grey lace over cream lobes, genuinely perforated; a healer, spans between zones 8 **NOUMEN** a pastel triangle-fractal plate; reachable only from long stability or from crisis Every cell carries a weight for each regime, plus built substrate, perforation, delivered metabolic energy, measured geometric strain, front activity, the age of its current tenant, an inherited colourway, and the identity of the nucleus that first claimed it. The dynamics that run over this are a real ecology, not a blend: **Competitive exclusion.** A regime is suppressed more by its rivals than by its own kind. This makes coexistence _locally unstable_ — a cell cannot hold a comfortable mixture of everything, and whichever regime is marginally ahead takes the ground outright. The skin is therefore a mosaic of committed territories with sharp fronts, rather than a uniform average of all nine. **Cyclic dominance.** The pairwise pressures between regimes are wired as a cycle, not a ranking. HISTION feeds on the builders; VELUM heals over HISTION; CORALLITE overgrows hard structure; REEFBLOOM cracks the crystal open; NOUMEN supersedes what has crystallised. There is no strongest regime, so dominance rotates indefinitely. **Niche construction with a lag.** Builders enrich the substrate until it is rich enough to support the parasites that consume them; the parasites strip it; the builders return to bare rock. The delay between those events is what makes the system oscillate rather than settle. **Spatial niches.** The plate offers three kinds of ground — face interiors, the thirty edge bands, the twelve vertex caps — and different regimes prefer different ones. Several can therefore hold territory permanently in their own zones while turnover happens within them. **Disturbance.** Wounds open periodically: a patch is cleared, its dead tissue becomes substrate, its stored energy is released, and fast pioneers that cannot survive under a closed canopy get somewhere to land. Old ground also loses its grip — a tenant that has held a cell a long time becomes easy to invade. The consequence is that the skin never arrives anywhere. It has no climax state. ![](https://storage.googleapis.com/papyrus_images/a46923971ed07a715bd2e3a5d84d2ddd1de88c7282fc2ec2b0de955f005ba44c.png) * * * **Morphodynamic patterning** ---------------------------- This is the centre of the organism, and the reason it is a lifeform rather than a display. Morphodynamic patterning means feedback in both directions: local pattern drives change in physical form, and the resulting change in form feeds back to reshape the pattern that produced it. One direction alone is decoration. Both directions together is a system that is its own cause. **Pattern deforms form.** The regimes' displacement fields are sampled by the ray-marcher itself and physically move the membrane. Spikes protrude, polyps swell into fluted rosettes, webs raise ridges, plates cut grooves, glands bulge. Each of these is a quasicrystal cosine sum over the six fivefold axes at its own frequency, with slowly drifting phason phases — long-range order without periodicity, which is why the surface reads as structured but never repeats. At the same time, the territorial shares pull on the **33-coefficient body plan** itself. This is the deeper coupling: when a regime takes ground, the organism does not merely get repainted. Its radius profile, its membrane thickness, its core, its spindles, its whole silhouette move toward what that regime is. Skin becomes body. **Form deforms pattern.** The organism carries a proprioceptive probe — a governor that samples its own distance field at several hundred fixed points and measures the geometry it _actually has_, including every deformation the skin has just made. Crucially, the probe's model of itself includes the skin: the same modes, the same carving, amplitudes read from the same cells. It is not measuring an idealised body. What the probe finds is local strain — places where the interface has become badly behaved, where the gradient of the field varies wildly. That strain is deposited back onto the very cells that caused it. And strain is an ecological variable: VELUM thrives on it, AURUM cracks under it, the conduit lattice routes energy toward it, fatigue accumulates from it, and above a threshold the governor overrides the organism's intentions entirely and forces it to resolve into a calmer form. So: the pattern makes the form, the form is sensed, and the sensation changes the pattern. Round and round, at every frame. ![](https://storage.googleapis.com/papyrus_images/5f7f10e2e1fe3ac60183ea67de3ebf7ca1ad38ff07e33df7bd61d9fd20180faf.png) * * * **Topology changes dynamics** ----------------------------- The skin does not only displace. Where openness and web amplitude coincide, it **carves** — opening real holes clean through the membrane. The organism's connectivity genuinely changes; you can see through it into its own lit interior. This is not a visual effect, because perforation is also an ecological parameter. Cells that have been opened gain up to 2.8× the dispersal mobility: patterns propagate through opened ground in a way they cannot through closed. A regime that opens the membrane is changing the diffusion geometry of the surface it lives on, which changes which regimes can reach where, which changes who opens the membrane next. Builders heal the holes closed again, and the connectivity changes back. Topology is therefore not a consequence of the dynamics. It is one of the variables. * * * **Metabolism** -------------- The organism runs an energy economy, and it is genuinely limited by it. The conduit lattice is the delivery system. Its twelve vertex nodes accumulate charge from a metabolic budget. Packets spawn at the most charged node and hop along the struts toward the **neediest** region — need being a compound of measured strain, unmet energy demand, and front activity. The animal feeds its wounds and its battle lines first. Where a packet arrives, it deposits energy into the cells around that vertex, and energy gates growth. The expensive regimes — the cyan machinery, the spiked bloom — simply cannot establish in a starving organism; the cheap, patient generalists win a poor world. Vivid tissue is expensive tissue, so the creature has to be able to afford its own extravagance. The grown lattice is the most expensive organ of all. It is a structure built _outside_ the body, and it decays unless the organism is running a surplus. Cheap to want, costly to keep. * * * **The endocrine layer** ----------------------- Underneath the ecology runs a slower, hormonal system: three coupled variables that behave like arousal, openness and tone. They drift under Ornstein–Uhlenbeck dynamics driven by fourteen incommensurate internal clocks, and they are pushed around by fatigue, by novelty hunger, by how much of the skin is currently at war, and by what the skin has actually become. Mood sets the organism's phase, and the phases are behavioural states: * **DWELL** — quiet; the form consolidates. * **WANDER** — restlessness; mutation temperature rises. * **INTENT** — the organism decides what to become, clears ground for it, seeds it, and withdraws support from the incumbent. This is a deliberate act of self-remodelling. * **SURGE** — an energy bloom; spindles extend, the quasicrystal deepens. * **COLLAPSE** — die-back of whatever currently dominates. * **CRISIS** — a mutation storm: the phason lattice lurches, and the body plan makes a discontinuous jump. Fatigue accumulates from strain and from the metabolic cost of holding open fronts, and it is relieved by resolution. Hunger accumulates when the trajectory stops moving. Neither is decorative: both feed back into which phase comes next. ![](https://storage.googleapis.com/papyrus_images/6a539917f46347a48bff9d580267ab72dc90c5662d12e9f9a3c7b665ba0b8a5d.jpg) * * * **It makes its own light** -------------------------- There is no key light, no fill, no rim, no sun anywhere in this work. The scene contains one object and a void. Every photon in the image is emitted by the creature itself, which means that what you can see of it is exactly what it is currently doing. Four sources, all internal: * the **twelve node lamps** of the conduit lattice, each burning the colour of the region it serves; * **energy packets in transit**, which are real moving point lights sliding along the struts; * the **core furnace**, which by geometry lights the inner walls seen through every perforation; * and the **skin**, lighting its own neighbourhood. The skin makes light three ways, mixed according to which regimes hold it. **Bioluminescence** runs slow travelling waves along ridges, rosettes and rims. **Fluorescence** glows channel-pure and structure-correlated, in the manner of a stained slide: one colour on the mass, another on the fibrous network weaving through its gaps, a third on nuclei that cluster denser inside the mass. **LED** is hard, quantised and blinking — live traces, flashing vias, panel lamps, no falloff at all. Each chemistry writes onto its own pattern, so territories keep their identity instead of summing to white. All of the patterning is generated in-shader from the organism's own symmetry — value noise, fractal Brownian motion, cellular fields that yield true radial fluting around each cell centre, honeycomb reticulation, a fabric-fine mesh for the lace regimes, fibrous ridged noise, nuclei speckles, and a rectilinear circuit generator mirrored by the plate's own symmetry. There is not one image file in the entire work. ![](https://storage.googleapis.com/papyrus_images/2dcf565f3aa71cfe96eb20250d2f334d8c6145fa286b5a5903b10fb178de8dd8.jpg) * * * **Navigating morphospace** -------------------------- An organism with a 33-dimensional space of possible bodies could still, in principle, spend its life in one small corner of it. This one cannot, because it is built to be permanently dissatisfied with any body it has already worn. The animal carries a **morphogenetic genome** — its own free coefficient vector, which the ecology modulates but does not dictate. Against that it runs a continuous search, and the search has no objective. Nothing in the organism is trying to be good, or fit, or optimal. Instead it measures **novelty**: how unlike everything it has previously been a candidate form is. An eleven-dimensional descriptor characterises what a morphology _is_ — its spikiness, its relief, which symmetry it actually expresses, its quasicrystal content, how open it is, whether it has a lattice, how much of it is core, its scale. A permanent archive of everything it has been supplies the comparison, and novelty is the mean distance to the nearest neighbours in that archive. Because the archive never forgets, a region of morphospace stays unrewarding once it has been occupied. The only way for the organism to score is to leave. This pressure cannot be satisfied by settling down, which is precisely the point. Alongside the archive it keeps a **map of niches** — a coarse grid over the descriptor space holding the best-resolving example it has found of each kind of form. Competition happens only _within_ a niche, never globally, so there is no single champion for the animal to collapse onto. The map lets it return to and recombine distant morphologies rather than merely drifting away from them. And it moves through the space with operators that cross ridges rather than slide down them: relocating a handful of coefficients outright, switching whole morphological systems on or off, recombining two distant remembered forms coefficient-by-coefficient. When it detects that its current form has stopped being novel, it becomes restless and deliberately leaves for somewhere it has not been. This is what multiscale competency looks like from outside: an animal that reorganises its own body as a way of exploring what bodies are possible, with the low-level ecology and the high-level intention both steering, and neither in charge. ![](https://storage.googleapis.com/papyrus_images/d74b984ad93d4a9f44c39124f76f24f971a6a3cfba5466718e2bd3783cc97105.png) * * * **Heredity** ------------ The ecology mutates as well as competes. Every new colony is founded by a **strain** — an archetype plus random deviation in its colourway, growth rate, dispersal and displacement mix, with the deviation scaled by the organism's current mutation temperature. Conquered cells inherit the colourway of whoever conquered them, so a successful strain's signature spreads across the skin with its territory. A strain that holds ground long enough — measured as an integral of territory over time — is **assimilated**: its deviations are written back into the live genome of its archetype, permanently. The regime itself has changed. The organism logs it, and its generation counter advances. So the nine regimes an hour-old organism is running are not quite the nine it was born with. * * * **Why no two are alike** ------------------------ Time in this work is advanced by the frame intervals the browser actually delivers. Not a fixed step, not a smoothed average — the real, jittering sequence of durations produced by this machine, this browser, this moment's load. For a non-linear continuous system that is not a phase offset. A different sequence of time steps produces a _different trajectory_, not a shifted position on the same one. The organism's adaptive controllers deepen this: they measure real rendering strain and feed the result back into the metabolic budget, so a machine under load genuinely starves the creature a little, and the adaptation becomes part of its history. Over long durations, floating-point divergence in the integration of non-linear dynamics does not stay as lag; it pushes the system onto neighbouring trajectories that never reconverge. The module ships with the demonstration. SETHIX.divergence(T) snapshots the organism's complete state, replays it under two different sequences of time steps, and returns the two resulting state hashes. They differ. Every time. Each instance also carries a **lineage hash** derived from its territorial weights and its coefficients — an identity that changes as it lives. Two organisms given the same seed are twins only at the instant of birth. Every canvas is its own animal, with its own history, and when the tab closes that individual is gone. * * * **Meeting it** -------------- Open the browser console. The organism is exposed on `window.SETHIX`. From there you can read its live state and occupancy, the event log, the morphospace descriptor and novelty score, and the niche map. You can isolate a single regime, inject a colony, provoke a crisis, run SETHIX.divergence(T), or step time forward. Leave the tab open and it continues — competing, opening, healing, feeding its wounds, resolving under strain, mutating, assimilating, moving through bodies it has not yet been. Come back in an hour and it will be something you have not seen. * * * ICOSA MORPH 2 ============= **Cladogenesis in a single file** ![](https://storage.googleapis.com/papyrus_images/85ed57ca2da434068e7e677d73c069ae79e1e101bd1d69cdc115028157597ce5.png) Transmutation XLV (ICOSA MORPH 2.1) [ **Transmutation XLV #45 - Substrate Transmutation \[SETHIX\] by Iain Ball | OpenSea** Pure WebGL SETHIX modules https://opensea.io ![Transmutation XLV #45 - Substrate Transmutation [SETHIX] by Iain Ball | OpenSea](https://storage.googleapis.com/papyrus_images/27465244cf5c0e22638e6a6aa3b8f2950b94da5a6277acff5cb8d75f5e44e2eb.png)](https://opensea.io/item/ethereum/0x71eb648bdd6ebb161dcda890dfadd1d28972ea42/45) _An organism that lives an hour as itself, is judged on how it lived, is granted read-only access to one of three siblings, and spends the rest of its life growing that inheritance through its own body — arriving nowhere, and never twice in the same shape._ ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- **ICOSA MORPH 2.1** is _Transmutation XLV_ in the pure WebGL SETHIX module series. One HTML file. Pure WebGL2. No dependencies, no external assets. A primitive digital lifeform whose unforked hour remains the ancestor, and whose later life is a clade grown through that ancestor rather than drawn over it. For the first hour it is the animal described in the companion piece — a body written in the icosahedron’s own coordinates, a skin of nine regimes competing across a geodesic lattice, a metabolism that feeds its wounds first, and light it makes itself. Then it speciates. ![](https://storage.googleapis.com/papyrus_images/1ca416a70a81c9bd4a6fe82f8ffbd3f19d9ac545e84441cff5f5e61487fcee91.png) ?seed=6b20cec7f4457ab405598d88ef60bfcd2e97538003df1878c7622c6dbf61c288 ![](https://storage.googleapis.com/papyrus_images/382299c0301080ab6825d7c3fcbdd0f18a270e0243fb9549ab256214525d495f.png) ?seed=03b6b71061267a70a843f6af3d8d2b1eb442e97c49692ef35750f5c71dc20b10 At one hour of accumulated frame time the organism is sorted, permanently, into one of three categories and granted read-only access to one of three sibling modules. Nothing about that is arbitrary. The category is _earned_: the organism has been measured, all hour, on four independent readings of how it actually lived — how far it travelled through its own space of possible bodies, how decisively it committed to any one of them, how particular rather than average it became, and whether its motion had real phrasing or only tremor. A quiet animal that held one locked, distinctive state for fifty minutes scores as highly as a restless one, and for opposite reasons. The verdict is written once and never revised, and there is no way back to being unforked. What it inherits is not a colour scheme. Each sibling is a whole organism with its own body plan, and it is that architecture which crosses: a rhombic cage of bowed ribs and swollen hubs, folded a hundred and twenty times through the icosahedral group; a thick bone shell with bevelled windows cut on the twelve fivefold and twenty threefold axes; a colony of branching tubes that sheds satellites and goes on governing them. The host does not become the sibling — no foreign code is ever run, and what crosses the boundary is a table of numbers. It grows the sibling's architecture _through_ its own, and its own quasicrystal, its own ecology, its own geometry remain the substrate that architecture is expressed in. Nor does it settle there. A clade is not a costume; it is a family of nine developmental states, and the organism keeps crossing between them — raising a configuration, holding it long enough to be read, leaving it, arriving at it again an hour later wearing a different material — for as long as the canvas is open. Two organisms given the same seed are identical for exactly one hour and never again. **And that hour is untouched.** Not "barely affected" — bit-identical. The proof is by construction: a variant is built with the entire 2.1 layer deleted outright, run against the real module from the same seed, and their state hashes compared at every checkpoint. They match. The same comparison against the pre-fork release build matches too, so a seed saved before any of this existed still opens the same animal. What follows is a description of what happens after that hour. * * * The Timeline Identidy, and what it is for ----------------------------------------- Before the first frame the organism is given 256 bits, fixed for the life of the canvas. The timeline ID reduces three independent sources with a self-contained SHA-256: the host's cryptographic random source; a _jitter probe_ that races the high-resolution clock against a busy loop twenty-four times and harvests the operating system scheduler's own indeterminacy; and a canvas fingerprint, in which a small 2D surface rasterises a gradient, a line of text, a composite operation and a curve, and the pixels are folded to thirty-two words — rasteriser rounding, font stack, colour management and GPU compositing all leaving device-specific residue. A salt unique to this module's lineage is mixed in last, so the same machine running a different SETHIX module cannot arrive at the same name. The first word becomes the simulation's seed, so the trajectory is an expression of the name rather than something running alongside it. Append the hash to the address as `?seed=` and that organism's first hour returns exactly. The ID has one further job. Everything the inheritance does stochastically — which trait families this lineage commits to and which stay plastic, the phases of its developmental clocks, where foreign expression first enters the skin — is drawn from a **separate stream** seeded from the name's later words. Nothing in the genetic layer ever touches the simulation's own random source. That separation is what makes the first hour provably inert, and it also means two organisms in the same category still hybridise at different speeds, in a different order, to different depths. The category names the parent. It does not name the child. ![](https://storage.googleapis.com/papyrus_images/a67fb4d3a2ca016cf93785410e45fe8767c42309321318971cd8086cb8c9fce3.png) seed=432ee65db5520c7d66d69356ddec9414cc489edeec4f07ebff310a703e84e5b0 * * * The verdict ----------- For the first hour the organism runs and is measured. Four independent things, on a fixed cadence of accumulated frame time so the reading is a property of the animal rather than of the machine's frame rate. **Intensity** — how far the body travels through its own 33-dimensional space, how much territory changes hands between the nine regimes, how often the endocrine phase turns over, how hard the governor works. **Decisiveness** — the opposite virtue. Occupancy entropy, sustained dominance, how long the leading regime holds, how settled the emitted colour stays. **Distinctness** — quasicrystalline clarity, lattice presence, clean perforation, and how far the body sits from the centroid of its own archetypes. Being _somewhere_, as against being in between everything. **Temporal character** — dwells and surges measured against the organism's own lifetime average speed, weighted toward persistence, because what separates phrasing from jitter is not that a state is different but that it is _held_. These combine so that intensity or decisiveness can carry a high score alone, with distinctness breaking the tie. A quiet, committed organism out-ranks a chaotic multi-skin cloud, which is the correct judgement and not the one an activity meter would make. The composite is mapped through percentile curves fitted to an ensemble of independent hour-long lives, so across many loads the three categories occur about equally often. Written once. Never revised. category inherits from **ACTIVE** SETHIX · LATTICE · 09 **NEUTRAL** SETHIX · BIO-ARC MORPH **DORMANT** SETHIX · COLONY MORPH ![](https://storage.googleapis.com/papyrus_images/a860f318518c34e540ac743115a51a008e35d3a2abd7ba7d9a1a58454931b4f1.png) **ACTIVE** ![](https://storage.googleapis.com/papyrus_images/92dc5015797eeb3c9361102c9b51958dd853d3392a5acbdb86b01f506e47ff6b.png) **NEUTRAL** ![](https://storage.googleapis.com/papyrus_images/f6c03d28ae7513457e1f9be80f2bd07bf2009e35a7a4a7051f9a9ff5ae826713.png) **DORMANT** One base per category. No trait crosses between them. **And no foreign code is ever run** — what crosses the boundary is a table of numbers, transcribed once from a reading of each module: measured palettes, patterning frequencies, structural radii, displacement amplitudes, dynamical rates, behavioural statistics. The organism re-expresses them through its own machinery. Nothing borrowed executes. * * * The fold -------- The three sibling modules have one thing in common, and it is not a look. All three build their bodies by taking a small number of explicit primitives and **folding them through the 5-3-2 spherical triangle** — the fundamental domain of the icosahedral group, order 120. LATTICE · 09 is one bowed Bézier rib running between a five-way hub and a three-way hub, at a rib radius of 8.6% of the circumsphere with hubs half as wide again, fused at a generous fillet — folded into thirty ribs and thirty-two hubs. Plus an inward spoke carrying a punctate bead, and a chain of swelling tubes growing from a seed at the exact centre, folded into a hundred and twenty. COLONY MORPH is five branching chains through the same fold. BIO-ARC is a thick shell with bevelled windows cut on the symmetry axes. The host, by contrast, draws _fields on a sphere_ — radial functions, barycentric networks, quasicrystal sums. That difference is not cosmetic. A field on a sphere can always be read as a finish; a folded rib cannot. So 2.1 gives the host the fold. The three mirrors are derived from the organism's **own** icosahedron, so the borrowed architecture lands on the symmetry the body has had since before the fork. Alongside them come the primitives the donors are built from: the bowed rib, the exact capsule segment, the smooth minimum, and a frontier-advanced chain whose radius swells on two scales of noise. The chain's control points are strictly positive blends of the domain's three corners. That places every one inside the spherical triangle _by construction_, and because the triangle is convex, so is every segment between two of them — which is what lets one tendril fold into a hundred and twenty without any of them escaping the domain and tearing the body open. The frontier does not saturate: it reaches out, withdraws a few segments, reaches again, and a fork channel retracts to nothing episodically and re-extends along whatever path the clock has moved its control points to. Growing a limb, resorbing it, growing a different one. * * * Authority: what the host gives up --------------------------------- Structures alone cannot speciate an organism. Hang poles, ropes and tentacles on an unchanged host and all three clades are the same animal wearing three sets of ornaments — its membrane still owns the mass, its geodesic cage still owns the outline, its spikes still own the detail, its palette still owns the light. So 2.1 carries a single scalar, **authority**, which is how much of the body plan the foreign architecture currently holds. It is spent _before a single foreign structure is drawn_, on the host's own primitives: * the membrane's radius and wall depth, * the width of the host's spindles, * and the presence of the host's geodesic cage. Every one of them keeps a floor. The host yields; it is not deleted. Its quasicrystal, its skin ecology and its lobing are half of what makes this a hybrid rather than a copy of a sibling, and they remain readable at the deepest expression any clade reaches. The cage in particular yields to authority rather than to any one trait. That is the difference between a clade that can leave its geodesic shell and one that wears the shell forever with something happening inside it. **Coherence is a floor against collapse; it is not a ceiling on morphology, and a cage that cannot be left is a ceiling.** A committed clade leaves it. A clade easing back toward its ancestor — which the residue states are — finds it waiting. * * * Topology, not relief -------------------- Relief, palettes and attached structures can all be read as decoration on an unchanged animal. **A hole cannot.** An outline with pieces missing from it is a different body plan, and it survives being blurred, which is the test that matters. Two of the three siblings do not have a closed surface at all, and the host learns both operations: **BIO-ARC** carries bevelled **windows on the symmetry axes** — twelve on the fivefold, twenty on the threefold — cut clean through a wall that thickens as they open. That the openings sit on the axes rather than scattered is the architectural tell: they are _ordered_, and the wall left between them reads as structure rather than as tatters. **LATTICE** raises a **filigree crust** — a reticulum standing proud of the membrane whose cells are then taken away, leaving a shell you can see into rather than a ball. Both are the same operation on the same primitive: raise the walls, then remove the cells. * * * Development is a program, not a blend ------------------------------------- A clade is not a costume, and it is not a slider. Each of the three carries a **repertoire of nine states**. A state names which of its structures are expressed, at what amplitude, how much authority the foreign body plan holds while it runs, and which material system the surface is made of. ACTIVE · LATTICE NEUTRAL · BIO-ARC DORMANT · COLONY residue residue residue coated pebbled framed reknit roped limbed lumen chamfer colony nano packed shed swarm spired caged filigree woven lobed hubbed granular bundle plum torus drift The organism moves between them on a slow clock, reorganising over about twenty-two seconds and then **holding** the new configuration for forty seconds to two minutes — long enough to be read, short enough to be watched. Selection prefers states it has visited least recently, never repeats in place, and actively prefers a state that changes the _material_ as well as the form, because colour is a system that travels and needs its own pressure to do so. Amplitude comes from the state. The trait families, the invasion field and the coherence governor decide how fast the organism moves through its repertoire, which states it can reach, and whether it must fall back to a cheaper one — they gate the **schedule** rather than scaling the result. This distinction is the whole architecture: a governor may slow a move or refuse one; it may not quietly multiply every structure in the body by two thirds. Every clade's `residue` state is a full member of the repertoire, not a starting point it leaves. The organism returns to it, and the ancestor is plainly visible again when it does. Maturity is a **clock**, not a product of conditions: the clade develops over roughly three and a half minutes of its own accumulated life. The invasion crossing the skin and the trait weights hold the _rate_; poor health stops it. Neither can scale down what has already been reached, and maturity is not re-earned every frame. * * * Four materials per clade ------------------------ Each state also names which of its donor's material systems is running. The systems are read from the siblings' own ranges rather than invented, and **the fourth in every clade is the host's own skin ecology**, carried through at reduced authority — the ancestor has to be reachable as a _material_, not only as a shape. **ACTIVE** walks the LATTICE albedo ladder from near-black plum through hot magenta to a pale bead pink; its bright end; a near-black liquid crystal drawing itself in schlieren, the nematic under crossed polars where the darkness _between_ the interference bands is what makes it birefringent rather than an oil slick; and the host's skin. **NEUTRAL** runs bone ceramic against saturated cobalt as separate cell-assigned domains; a **thin-film interference skin**, a whole rainbow across a pale shell whose hue is set by film thickness rather than pigment, the bands following the geometry the way a fringe does; a warm mineral reticulum, orange-red network over a dark ground; and the host's skin. **DORMANT** runs bold exclusive dye pairs — lime against magenta, purple against green, teal against orange, a gold particle field — each normalised against the winner and raised to a hard power so one label holds one place. In an acquisition most of the field is black: what is not labelled is not dim, it is absent. Underneath, each donor's own **palette epoch** turns slowly — COLONY carries a fluorophore epoch and a stain-matrix rotation, BIO-ARC a palette table on a 92-second clock — so which dye has bound to which structure, or which substance holds the ground, changes over minutes as well as between states. **And the host's light recedes.** The host lights itself from twelve node lamps, packets in transit and a core furnace. That rig is the _host's_ material language, and left at full strength under a borrowed finish it flattens three particular materials into one bright hue. LATTICE dims it hard, because a lamp on a crest is the one thing that cannot happen in an electron micrograph. BIO-ARC _raises_ it, because bone does not glow, it is seen, and its two substances separate only under actual light. COLONY dims it almost to nothing, because there is no illumination in a confocal image at all — every term is emission and occlusion acts as absorption. Emission added on top of a shaded surface is a glow over the whole body; emission that modulates the albedo is a material, and the clades want the second. * * * Hybridisation is a place ------------------------ The inheritance does not arrive as a switch and does not arrive everywhere at once. Foreign expression enters the skin at permanent entry sites chosen by the organism's name and spreads across the geodesic lattice as a **logistic travelling wave** — a real invasion, with a carrying capacity set by how much the organism has agreed to take on, and a growth rate opened up by local strain. Hybridisation follows the places the body is already working hardest. So there is a front. On one side the cells run the original genome; on the other the hybrid; and the front moves. In the shader it is lit by **both parents at once** — the organism's own accent against the foreign structure colour on a slow rotator, with a white core where the gradient is steepest. It burns hottest while the outcome is genuinely contested, fades as either genome takes the ground, and steps back entirely for a clade that has finished arriving, because a front is the act of conversion and not a decoration. Six trait families cross independently and at different rates — skin, colour, perforation, displacement basis, attractor bias, metabolic tempo. Colour crosses first and furthest, because a hue cannot tear a membrane. The displacement basis crosses last and least, because it is the one thing that can actually break the body. ![](https://storage.googleapis.com/papyrus_images/c776d34ca5671689019a5175e167f4035167b1f37d206b612323e65aa6afcd9a.png) ![](https://storage.googleapis.com/papyrus_images/a2d52e81ae225e37cac054790e93b4d1105a85091116202ddbd618a4c33d62da.png) * * * Keeping the trait space open ---------------------------- An engine that only advances arrives somewhere and stops. Three mechanisms keep this one moving. **Fixation, per family.** What has been carried safely for long enough stops being provisional: a floor creeps up under each trait family and the engine may not propose below it. But the strength of that ratchet is drawn per canvas, so which characters _this_ lineage commits to and which stay plastic is part of its identity. And it leaks — slowly, even in perfect health. Nothing is ever locked; it is only expensive to move. **One exception.** The family paying for a clade's diagnostic structure — the thing whose absence would stop the organism being a member of the clade at all — ratchets rather than tracking, leaks an order of magnitude more slowly, and is priced at half. A clade's defining character is the thing it exists to express; pricing it as a luxury is how a lineage wanders out of its own clade. **Trait-range occupancy.** The novelty archive measures where the _body_ has been. A second, coarse, decaying histogram measures where the _inheritance_ has been — how long each family has spent in each part of its own range. A proposal that moves a family somewhere it has rarely been is worth taking on those grounds alone. This is coverage pressure rather than a drive toward the ceiling, and it is why the organism travels between ancestor and sibling instead of running to the sibling and parking. Retreat also becomes likelier the further into its range a family already sits, so an established inheritance is genuinely revisited rather than merely defended. * * * What stops it becoming a monster -------------------------------- Five governors, none of which can be switched off, and all of which are floors rather than ceilings. **A structural integrity bound.** No foreign influence may displace a coefficient beyond a bounded fraction of that coefficient's own range, and none may move faster than a fixed rate per second. The organism can be pulled. It cannot be replaced. **A coherence metric.** Five readings from the _actual probed geometry_ — surface area, solid fraction, strain, gradient magnitude, and whether the body is fighting its own normalisation — combined as a **geometric mean**, so any single collapse drags the whole reading down. A body can be perfectly reasonable in four respects and still be a cloud of dust in the fifth. Below a safe threshold the engine reverses its most recent borrowings, newest first; below a critical one it stops acquiring and withdraws. **Two clocks on that metric.** The reversal governor reads a fast signal, smoothed over about three seconds, which is right for rejecting single-frame probe noise. The _structure_ subsystem reads a slow, asymmetric one — falling over forty seconds, recovering in nine. The reason is that the host runs strain excursions entirely of its own, with nothing inherited involved, and on a three-second signal those transients reach in and withdraw a structure that took minutes to grow and had nothing to do with causing them. Slow to abandon an inheritance, quick to resume once the body is well. A sustained loss still takes everything back. **A brake on arrival.** While coherence is falling, structures build more slowly. This is a brake on the _rate_ and on nothing else — the ceiling, the repertoire and the palette are untouched, so the clade still reaches every state it could reach; it simply does not slam a new body plan into a body still struggling to hold the last one. And when a state the organism is holding rises above what the body can currently support, the dwell is cut short and it leaves for a cheaper one. It moves; it does not stop expressing. **A novelty-against-stability trade.** Every proposed incorporation is scored on both axes. Novelty counts distance from where the organism has been and from its phantom siblings; distance from the _pure foreign morphology_, because becoming the other thing is imitation and not novelty; and **particularity** — a bonus for rare, coherent, decisive states. Persistent chaotic competition of many skins is the commonest thing this system can do and is therefore scored as the least novel thing it can do. **A metabolic gate.** Borrowing is permitted in proportion to present health — energy reserve, governor load, fatigue. A strained organism slows and spends its time reversing instead. Measured across eighteen independent post-fork lives in all three categories: no non-finite values, no fragmentation, no fusion to solid, the distance field marchable throughout, and worst-case coherence comfortably above the critical threshold with the reversal governor idle. **And it is completely reversible.** Raise the coherence floor above anything the body can reach and hold it there: every trait weight, every fixation floor, the coefficient offsets, the invasion field and every inherited structure return to exactly zero, and the original organism comes back. Release the floor and it takes its inheritance up again. * * * What you actually see --------------------- The tempo is the point. An organism whose development is only visible in time-lapse is holding still as far as anyone watching is concerned. A state crosses in about twenty seconds and holds for one to two minutes. A clade matures in three and a half. Its palette epoch turns in four. Authority — how much of the body the sibling currently holds — wanders on its own clock of about six minutes, shaped to hold rather than sweep, so the organism commits, stays committed, eases decisively back toward its ancestor, and returns. Within twenty-five minutes a clade has visited all nine of its states. So an **ACTIVE** organism is a dark host mass in a geodesic cage; then a lobed body with a gold crust arriving over it; then the cage genuinely re-knitting into a rhombic clathrin lattice with large openings and swollen hubs; then that lattice with its lumen filled by a folded nano-colony threading a packed bead mass, lit gold from inside; then a pale filigree with poles crossing an open interior; then back toward a magenta ancestor with spikes. A **NEUTRAL** organism is a pale lobed mass; then a bone star; then a copper mineral star; then a beaded rope cage with real apertures; then an iridescent oil-slick shell whose rainbow follows its own curvature; then a granular packed interior behind bone. A **DORMANT** organism is a grey ancestral star; then a warm quasicrystalline host skin; then a fine dodecahedral frame around a colony; then dense radial tube bundles in lime and magenta with pods shedding off it; then teal and orange with satellites orbiting on the organism's own fivefold axes, sized and governed by the parent that let them go. None of it is a costume, because none of it is a fixed destination. A clade identity is a family of states, and the organism goes on moving through the family. * * * Why no two are alike -------------------- Time here is advanced by the frame intervals the browser actually delivers — not a fixed step, not a smoothed average, but the real jittering sequence produced by this machine at this moment. For a non-linear continuous system that is not a phase offset. A different sequence of time steps produces a _different trajectory_, not a shifted position on the same one. The organism's adaptive controllers deepen it: they measure real rendering strain and feed it back into the metabolic budget, so a machine under load genuinely starves the creature a little and the adaptation becomes part of its history. `SETHIX.divergence(T)` ships the demonstration — same state, two sequences of time steps, two different state hashes. After the fork there are four further sources of divergence. The name differs, so the seven imagined siblings the novelty drive steers away from differ. The rates, ceilings and fixation strengths at which the six trait families cross differ. The phases of the developmental and palette clocks differ. And the order in which the nine states are visited differs, because selection depends on what the organism has recently done. Two organisms given the same seed are twins at the instant of birth and nowhere after it. ![](https://storage.googleapis.com/papyrus_images/70b345a104386af7792b221ad6ed31966874035b977fdb3c1fee0e11b890a8d8.png) * * * Meeting it ---------- Bottom right of the canvas: the label ICOSA MORPH 2.1. It shows for two seconds after the first frame, then again when you hover that corner or tap it. Click the label to open the card. Before the fork the card shows a countdown to the evolutionary split, in accumulated rendered time (a backgrounded tab pauses with the organism). After the fork it shows the permanent category and the genetic base that was granted. Always, below that, the 256-bit timeline identity — selectable, click-to-copy. To keep an organism, save that hash. Reload the module with ?seed= plus the hash and the first hour of that lineage returns. The same console object as 2.0 remains on window.SETHIX: live state, occupancy, event log, morphospace, niche map, isolate a regime, inject a colony, provoke a crisis, force the fork early, step time. Leave it running. It lives an hour as itself, then forks, and spends the rest of its life growing a sibling’s architecture through its own body — holding a configuration long enough to be seen, and leaving it for another. * * * All code is open source. Console: window.SETHIX. Contract: `0x71eb648bdd6ebb161dcda890dfadd1d28972ea42` — [**Substrate Transmutation \[SETHIX\] on OpenSea**](https://opensea.io/collection/substrate-transmutation-sethix-515117031). Media on Arweave. * * * --- *Originally published on [Iain Ball](https://paragraph.com/@iainball/computational-cladogenesis-from-complex-cybernetic-organisms-to-primitive-digital-lifeforms)*