The Natural History of Civilisations

Part I · The essay

The Human Metasystem

A Natural History of Civilisation

By Kari McKern

Imagine an alien naturalist arriving on Earth. She observes a peculiar primate. Its members wear coloured uniforms, salute pieces of cloth, kneel before symbols, and sometimes willingly die for people they have never met. They inherit obligations from the dead and undertake projects whose beneficiaries have yet to be born, explaining it through extraordinary stories: patriotism, religion, liberty, justice.

She asks not which stories are true, but a more elementary question. What kind of organism behaves like this? Not the individual human animal, but the larger entities that recruit these animals into coordinated activity, survive their deaths and adapt through generations. She has discovered the human metasystem. And she is about to discover that what humans call history is part of natural history.

1The Evolutionary Problem

Maintaining an organised structure requires energy: cells must repair, organisms must feed, populations must reproduce. Life persists by capturing useful energy and matter, maintaining local order and exporting entropy.

An ant colony shows how far cooperation can go: through simple signals its members build structures, manage resources and defend territory, with capabilities no single ant understands. Cooperation makes new forms of adaptation possible: the group becomes a functional entity in its own right, a new level at which capabilities accumulate, compete and persist.

Humanity inherited and transformed this capacity: tools, language, teaching, and eventually symbolic systems preserving information outside individual brains. Institutions made relationships durable beyond their participants' lives. The decisive transition was not greater intelligence, but that cooperation could accumulate knowledge beyond the individual lifespan. The group acquired a memory and a machinery for reproducing itself. In time it could build cities, maintain empires and send instruments beyond the solar system.

2Eight Things an Organised System Must Do

I began this investigation in 2024, approaching societies as a naturalist rather than a political theorist. Suppose nations really are complex adaptive systems: what would their essential functions be, and could their relationships explain their behaviour? The question led to the Complex Adaptive Model State, CAMS, and its eight recurring functions, the Paladins.

Every durable human organisation must establish direction. That is Helm. It must protect its boundaries and enforce collective decisions. That is Shield. It must preserve shared understandings, explanations and legitimacy. That is Lore. It must retain usable records, procedures and institutional experience. That is Archive. It must develop and reproduce specialised skills. That is Craft. It must perform the work upon which the organisation depends. That is Hands. It must circulate goods, information and resources. That is Flow. And it must allocate, maintain and conserve the assets necessary for continued existence. That is Stewards.

These are not departments or groups of people but functional requirements that must somehow be met. A deeper implication emerged: perhaps these functions are not merely a convenient classification but attractors around which cooperative organisation develops.

3The Attractors That Recruit Us

We arrive with inherited dispositions, emotional needs and an extraordinary capacity for social identification; organisations offer belonging, achievement and meaning, and the fit is never perfect.

Yet a remarkable regularity emerges. Societies continually recruit people into the same broad classes of cooperative activity, and participation reshapes them. Craft develops apprenticeships and professional pride. Archive develops recordkeeping and respect for continuity. Shield develops discipline and codes governing the legitimate use of force. Lore develops rituals and systems of belief. The institutions reproduce not only skills but motivations and identities.

The metasystem makes its functional requirements personally meaningful to the animals performing them: an important part, I suggest, of the origin of social complexity. The Paladin attractors are not personality stereotypes but recurring opportunities and pressures within the ecology of cooperation.

4How a Civilisation Grows Its Culture

Return to our naturalist. She visits ancient Egypt, a civilisation organised around a river whose flooding sustains agriculture. Stewards manage surpluses, Archive preserves records, Craft develops irrigation, Helm organises the effort. And Lore gives the arrangement meaning: the river's cycle becomes entangled with cosmology, kingship and pharaonic authority. The god-king is no eccentric belief: his sacred authority organises the activities the civilisation depends on.

On the Eurasian steppe, geography favours mobility; herds, kinship networks, portable wealth and rapid decisions create different demands, and authority may rest on military competence, allegiance and ancestry. The culture is different because the adaptive problem is different; a Thai river community, a Himalayan religious polity, a maritime republic and a modern industrial nation orbit the same requirements along trajectories set by geography, inheritance, energy and technology.

Culture is what these configurations look like when expressed through human imagination: the visible flowering of cooperative organisation in procession, uniform, code and flag. Flowers recruit pollinators. Cultures recruit participants. They attach affection, dignity, loyalty and purpose to behaviours that sustain the larger system. This is why flags matter. A flag is not an energy source, but it can coordinate people who command enormous resources. The underlying mechanism is not moral but organisational. Culture makes cooperation meaningful, and meaning changes what cooperation can accomplish.

5Meaning and Metabolism

An organised society is a material system: it consumes energy, transforms resources and exports entropy. An individual cannot construct a hydroelectric dam; a coordinated society can redirect a river to sustain an industrial landscape.

The resulting order is extraordinarily costly, paid in dissipation, mistakes and conflict. Yet people do not work to increase entropy production. They go because they need wages, love their families, enjoy making things or believe in a collective purpose.

Here is the remarkable coupling: the larger system's metabolism depends upon human behaviour, while symbolic meaning helps organise that behaviour. One cannot explain a symphony by measuring its electricity consumption, but no symphony exists without instruments, musicians, knowledge and sustained resources. Meaning and metabolism are aspects of the same process: circular, adaptive and self-reinforcing.

What looks like history from the human perspective becomes, from the naturalist's, the evolution of ever more elaborate dissipative structures. Civilisation does not escape thermodynamics through culture; it employs culture as one of the means by which its material organisation becomes possible.

6The Personality of a Nation

If every society performs the same eight functions but distributes its capabilities differently, might the resulting arrangement possess something analogous to a personality?

We routinely call nations cautious, inventive, militaristic or bureaucratic. CAMS offers another possibility: national personality may be an observable, persistent pattern in the relationships between functional systems. A society dominated by Shield and Helm behaves differently from one whose capabilities reside in Craft, Flow and Archive. None of this requires imagining a nation with a single brain; cognition, memory and capability are distributed across participants and institutions. The metasystem thinks, in this functional sense, through its relationships.

CAMS describes each function through four dimensions: Coherence, the alignment of activity; Capacity, the ability to perform; Stress, internally generated strain; and Abstraction, the integration of symbolic understanding with practical activity: 32 measurements in all. A more recent development simplifies the picture: take Abstraction multiplied by Coherence as cognitive integration and Capacity minus Stress as available functional margin, and multiply them:

σ = (Abstraction × Coherence) × (Capacity − Stress).

This yields eight values describing the relative activation and constraint of the functions. A society's personality is not the temperament of its average citizen but the character of the cooperative arrangement in which citizens participate. And because those relationships change, the personality has a history.

7Reading the Organism's Footprints

If CAMS describes real features of cooperative organisation, they should leave evidence, and societies leave astonishing quantities: decrees, soldiers' letters, merchants' complaints, administrators' accounts of institutions functioning splendidly, sometimes when they plainly are not. The historical record is the accumulated residue of an adaptive organism going about its business; every budget, regulation and dispute carries information about the relationships between functions.

Three observations recurred. High-amplitude signals, because major changes in coordination alter institutions, language and power. Consistent readings, because different investigators encounter the same conspicuous relationships. Consistent stories, because similar functional configurations produce recognisably similar behaviour in different circumstances. A dysfunctional company and a declining empire are not the same entity, but each can suffer failed leadership, damaged institutional memory or broken coordination. The historical costumes change. The functional relationships remain recognisable.

8The Surprise in the Experiments

Over two years CAMS grew from an exploratory model into a substantial research programme, revised through adversarial testing; the functional grammar kept generating recognisable patterns.

As the operating capabilities of Evergrande, a major Chinese property enterprise, deteriorated, its configuration resembled an enterprise whose productive life had been displaced by administrative management of its surviving obligations: Archive became the last prominent function of a dying corporate organisation. Vanke produced a different portrait: management authority challenged while operational and stewardship functions held the organisation together.

Could AI systems, given only compact numerical descriptions of the eight functions, produce portraits that other systems could match to recognisable historical eras? The November 2026 round covered 126 eras across core, extension and sparse-control sets. On the 56-era core panel, Claude and Gemini each matched every portrait to its era in the closed-list identification task; Kimi recognised 52; Manus also returned a perfect core score, but under a contamination flag. Grok's 19 of 56 still cleared the pre-registered bar; Codex sat at chance. Recognition weakened on the extension panel (Claude 45 of 51, Manus 16, Kimi 8) and sparse Denmark (Gemini 17 of 19, Claude 13, Kimi 12). A second arm, testing mood continuity against sealed historical grades, met its majority bar on the core set for both returning models, with no overturning disagreement. The audit kept the imperfections visible. Gemini's claimed hashes were typed, not computed, so its numbers are descriptive only. Claude's Arm A bytes are unverified. The pre-registered nulls were never returned; some cells were never run. The results are a promising demonstration of structured recognition, not a completed external validation or proof of the mechanism.

The deeper observation: a small functional description can carry enough structure to generate a recognisable account of an era. Russia's Thaw, Depression-era America, Whitlam's Australia and pandemic-era China differ enormously, yet can be described through the same functional language. The zeitgeist historians recognise as an era may arise partly from the changing relationships of its cooperative functions.

9Human History Rejoins Natural History

The divisions between biology, economics, history, sociology and politics begin to look less fundamental: different observational approaches to related processes of cooperative adaptation. A cell preserves information and allocates resources; an animal group achieves capabilities through collective behaviour; human beings build institutions that preserve memory across generations. A civilisation is not literally a giant cell, nor a company an animal. But at successive levels of organisation, relationships between components make possible capabilities no component possesses separately.

With humanity, those relationships acquire an extraordinary symbolic dimension: we can imagine organisations that do not yet exist and deliberately alter our arrangements. Human history is neither an exception to evolutionary dynamics nor a repetition of animal behaviour. It is the continuing evolution of cooperative organisation through increasingly sophisticated mechanisms of memory and cognition. There is no need to subtract love, beauty, loyalty, creativity or suffering from this account: they are among the experiences through which the participants live the process.

10And Then We Invented Artificial Intelligence

There is a final recursion in this story. Writing, printing and the internet transformed knowledge preservation and transmission. Now artificial intelligence is transforming the production and circulation of cognitive work. Lore, Archive and Craft are being altered, Helm gains new instruments, Flow accelerates, Hands encounters automation: not merely new tools within existing institutions, but changes to the relationships between functions, enabling new cooperative arrangements and new forms of dependency, manipulation and concentrated power.

The human metasystem is changing its own anatomy.

And humanity's accumulated knowledge has produced machines able to help us investigate the processes through which that knowledge was accumulated. My own research has emerged through such a collaboration: an exploration of cooperative intelligence conducted through cooperative intelligence. The subject and the method have begun to resemble each other.

11The View From the Moon

She watches a city at night. Electricity flows, hospitals maintain life, children learn the discoveries of generations. Millions perform specialised tasks without knowing one another; no individual understands the arrangement, yet the city persists. Its organisation exists not in any ruler nor in its inhabitants' intentions but in the relationships between them.

Their histories are written in stone and steel, paper and silicon, poetry and law, love and warfare; their cultures are the extraordinary flowers of their organisation. And their continued existence depends upon the familiar predicament of all living systems: preserving order in a universe where that always has a cost.

The surprise of these two years was not merely that eight functions could describe societies, but that their changing relationships generate recognisable signatures of history, recoverable across societies and organisations: a common underlying architecture of cooperative life. Human beings are biological organisms participating in larger adaptive systems, recruited through functional attractors, organised through culture and meaning. CAMS is an attempt to make those trajectories visible.

Its most important consequence may be not what it tells us about empires, but what it allows us to recognise in ourselves. We are not isolated individuals pursuing private destinies, nor helpless components of institutions beyond our comprehension. We are participants in extraordinarily elaborate cooperative arrangements, inheritors of accumulated adaptive knowledge, agents capable of changing those relationships. The species that learned to cooperate in hunting bands has built a planetary civilisation capable of examining its own organisational condition, and is developing, for the first time, instruments that could make the anatomy of our collective existence visible to the beings who constitute it.

The naturalist looking down from the Moon sees an astonishing animal. The animal is us. The larger organism is what we make together.

And its history is still being written.

Part II · The supporting document

The Human Metasystem: Supporting Document

Linking the Essay to the CAMS Research Programme and Placing the November 2026 Findings in Context

Kari Freyr McKern, Neural Nations Research, Sydney

Prepared 8 November 2026

Provenance: this document accompanies the condensed essay "The Human Metasystem: A Natural History of Civilisation" (2,299 words). The experimental figures cited here are taken from the sealed final report of the independent AI replication round closed on 7 November 2026, deposited in the KaliBond/wintermute repository under analysis/cams_personality_independent_ai_replication_2026-11-07/, together with its evidence seals, score packs and verify_evidence.py. Where this document compresses a caveat, the seals govern.

1What the Essay Claims, and on What Evidence

The essay is the public face of a research programme, not the programme itself. Its naturalist is a rhetorical device, but the discipline behind her is real: the Complex Adaptive Model State, CAMS, treats a society as a system of eight recurrent functional requirements, the Paladins. Helm establishes direction, Shield protects boundaries and enforces decisions, Lore preserves shared understanding and legitimacy, Archive retains usable records, Craft develops specialised skill, Hands performs the work, Flow circulates goods and information, Stewards allocates and conserves assets. Each function is scored on four vitals: Coherence, Capacity, Stress and Abstraction, giving 32 measurements per society per era. The compact personality operator used in the essay, σ = (Abstraction × Coherence) × (Capacity − Stress), is the cognitive integration of a function multiplied by its available functional margin, and yields the eight-number portrait from which the November experiments were run.

Three commitments of the framework are visible in the essay and should be read as load-bearing, not decorative. First, the eight functions are claimed as functional requirements, and tentatively as attractors, not as departments or classes of people. Second, culture is treated as the mechanism that recruits human motivation into functional roles; the flag coordinates, the story binds. Third, the coupling of meaning and metabolism is explicit: civilisation does not escape thermodynamics through culture, it employs culture to make its material organisation possible. Each commitment has a formal statement elsewhere in the programme: the first in the Compression Theorem, which holds the eight-node partition to be a regime-dependent coarse-grained basis rather than a universal upper bound; the second in the Extended Social Cognition Hypothesis, ESCH, which reads institutions as exteriorised functions of distributed social cognition; the third in the programme's dissipative-structure framing of social order as costly, entropy-exporting organisation.

2The November 2026 Replication Round

The question under test was narrow and pre-registered: can eight CAMS node functions, read only through Cog = A×C, Aff = K−S and σ = Cog×Aff under the compact pocket rules, produce era personalities recognisable against sealed historical eras, when written by AI systems that never see the original analyst pockets or mood grades. The grid covered 126 eras: a 56-era core panel (Russia 23, Australia 18, USA 15), a 51-era extension panel (China 20, UK 16, Norway 15) and a 19-era sparse-input control (Denmark, era IX unobserved). Six models participated: Claude (claude-sonnet-4-6), Gemini (gemini-2.5-pro), Kimi K2.5, Grok-4.7, Codex (GPT-6) and Manus.

The results, stated with their caveats attached, are these. On blind era identification from unlabelled pockets against a closed list, Claude and Gemini each scored 56 of 56 on the core panel; Kimi scored 52; Manus also returned 56 of 56 but under a contamination flag, since Arm A had read the tables in the same session; Grok scored 19 of 56, which still cleared the pre-registered bar of twice the mean row chance (10.71%); Codex scored 3 of 56, at chance, and failed the bar. Off the core panel, recognition weakened sharply: on the extension, Claude 45 of 51, Manus 16, Kimi 8, Codex 3; on sparse Denmark, Gemini 17 of 19, Claude 13, Kimi 12. The second arm, comparing pre-sealed pockets against sealed historical mood grades, met its majority exact-or-adjacent bar on the core set for the two models that returned (Claude 48 of 56, Gemini 46 of 56, each 27 exact), with zero overturning disagreements; but both returns concentrated on the two commonest sealed grades, which makes the bar a coarse screen, and Denmark did not look like the core: Claude's bare majority there sat beside four rows graded on the wrong side of a sealed disagreement.

The audit findings are part of the result, not footnotes to it. Gemini admitted that its claimed Arm A and Arm C hashes were typed as text rather than computed over the bytes, so its numbers are descriptive only. Claude's Arm A file was never received, and the tester times on its rows are later than the seal times; its Arm C output hashes match the sealed files, but the pocket lineage is unverified. The pre-registered nulls (shuffled σ, random means, a Wikipedia-only condition) were never returned. Arm A's 60 percent leadership-match criterion was never computed. Several cells were never run for resource reasons. The report's own conclusion is the one the essay now carries: a promising and repeatable demonstration of structured recognition, not a completed external validation, not a causal test of the theory, and not a validation of an organology.

3The Wider Validation Record, Including Its Failures

The November round sits inside a longer validation programme whose failures matter as much as its passes. On the positive side: a blind morphology-identification study recovered the exact identity of anonymised societies in 13 of 24 cases (P ≈ 1.2×10⁻⁶) and the functional family in all 24; the Relational Inelasticity Hypothesis, tested blind against an untouched 12-society holdout, returned Spearman ρ = 0.7017 (p ≈ 0.000025 over 40,320 permutations); leave-one-society-out structural-collapse prediction reached AUC 0.740 across 29 societies. On the debit side, and logged with the same discipline: the Stress Entrainment Hypothesis failed its pre-registered primary decision rule in October 2026, and σ was demoted from candidate operator to compact description, which is precisely the status the essay gives it; the Abstraction Rent Hypothesis failed its controlled-effect test; a peer audit of Persistent Shear found a critical circularity problem in LLM-scored data and the claims were revised down; the T-M series archive audit found mislabelled episode counts and non-reproducing figures, corrected through a numbered trail with verdicts unchanged. The programme's rule is that FAIL is FAIL, that deviations are logged, and that identical cells across models count as one instrument, not five witnesses. That last rule bears directly on how the November result should be read: the models share training corpora and historiographic conventions, so their agreement on the core panel is evidence of a reproducible signal in the pockets, not yet evidence of independent confirmation.

4The Complexity Sciences Context

The essay's claim that human history rejoins natural history places CAMS deliberately within an existing lineage, and the debts should be named. From thermodynamics, the framing of organised systems as dissipative structures, maintaining local order by exporting entropy, is Prigogine's; the essay's insistence that no symphony exists without instruments, musicians and sustained resources is that insight applied to civilisation. From the Santa Fe tradition of complex adaptive systems (Holland, Gell-Mann and others) comes the central move: agents with simple rules aggregate into entities with capabilities no component possesses, and the interesting science lives at the level of the aggregate. From evolutionary biology, the programme draws on multilevel selection (Sober and Wilson) and on the major-transitions literature (Maynard Smith and Szathmáry): the essay's claim that the decisive transition was not greater intelligence but the accumulation of cooperation beyond the individual lifespan is a major-transitions claim, with institutions cast as the new inheritance system. The ant colony is present for the same reason it is present in Hölldobler and Wilson: as the existence proof that coordination can produce a functional entity without central understanding. From resilience ecology, the programme imports Holling's adaptive cycle as its lifecycle anchor, with the conviction that the danger to a society is rigidity, not weakness. From distributed cognition (Hutchins) and the extended-mind literature (Clark and Chalmers), ESCH takes its warrant for reading institutions as exteriorised cognitive function; a metasystem that "thinks through its relationships" is an extended-cognition claim, not a mystical one. From cliodynamics (Turchin and colleagues) comes the demonstration that structural-demographic variables can be measured across centuries and tested, though CAMS differs in what it measures: not elite overproduction and state fiscal health, but the relational configuration of eight functional systems.

Where CAMS attempts a contribution of its own is in three places. It offers a fixed functional grammar, the same eight requirements applied identically to firms, nations and historical civilisations, which makes cross-level comparison a measurement rather than an analogy. It provides a compact relational state, 32 vitals reducible to an eight-number portrait, shown in November 2026 to carry enough structure for blind era recognition on well-documented core societies. And it supplies a measurement procedure with an audit culture: pre-registered bars, sealed keys, hash-verified blinding, published failures.

5The History and Philosophy of Science Context

The programme should be located honestly in the longer history of attempts to make history a science, because that history is mostly a record of instructive failure. The nineteenth century produced social physics in Comte and Quetelet; the twentieth produced the grand morphologies of Spengler and Toynbee, and the mathematising ambition of the Annales school and Braudel's longue durée. The morphologists found patterns but could not measure them; the Annales school measured but largely declined to theorise the whole system. CAMS is best understood as an attempt to do both at once, made possible now for a specific, nameable reason: the arrival of a population of large language models able to read heterogeneous historical prose and score it against a fixed rubric, at a scale no lone researcher could previously attempt. The multi-model ensemble is the programme's instrument, and the November round is best read as instrument characterisation: the first systematic test of what the instrument can and cannot recover.

That reading determines the epistemological posture. The programme follows Popper and Lakatos in treating pre-registered falsification as the price of admission, and its public record of failed predictions is the Lakatosian test of whether it is a progressive or a degenerating research programme; the stress-entrainment and abstraction-rent failures, and the refuted morphospace prediction reported in the blind morphology study, are kept visible for exactly this reason. From Kuhn it takes the warning that a framework that cannot be surprised by its data is not yet an instrument, a rule the programme states in its own epistemological bedrock. From the Duhem-Quine problem it takes the discipline of audit: when a test fails, the failure may live in the scoring protocol, the model ensemble or the theory, and the correction logs exist to keep those possibilities separable. From the philosophy of measurement it takes the construct-validity question as central: whether Coherence as scored conflates compliance, coordination and commitment is a named open problem, and apparent model failure is treated as potential evidence about the construct rather than noise. And from the historiography of science it takes a modesty about provenance: the framework's seed is the librarian's retrieval lesson, that a badly framed query returns your own assumptions while a well-framed one surprises you, and the whole validation apparatus is that lesson institutionalised.

There is also a deliberate Copernican move in the essay's vantage. The alien naturalist demotes polities to instances in a population: no chosen peoples, no teleology, no civilisation exempt from the functional requirements. This is the programme's declared politics of method. The same eight questions are asked of Washington, Beijing, Moscow and Copenhagen, and the scoring standard is identical.

6Where the Claims Stand

Stated plainly, the evidence today supports the following. A compact eight-function description of an era carries enough structure for independent AI systems to recover the era from a closed list at rates far above chance on well-documented core societies, with recognition degrading where input is sparse or historiography thinner, which is what a real signal should do. The same portrait shows descriptive agreement with sealed historical mood grades on the core set, with the caveats of grade concentration and one binding hash-fabrication admission. The framework has survived several pre-registered tests and failed others, and both classes of outcome are in the public record.

The evidence does not yet support: a clean external validation anchored outside scorer-generated data; any causal claim about the mechanism of social coordination; any generalisation to cells that were never run; or the strong reading of the essay's closing suggestion that the zeitgeist of an era arises from the relationships of its cooperative functions. That suggestion is a hypothesis with a promising instrument behind it, no more. The next steps are the ones the validation protocol already names: returned null models, an external anchor for the relational grammar, and permutation-null statistics attached to every headline before peer review.

The essay ends with the view from the Moon. This document exists so that a reader who looks down beside the naturalist can see the instrument panel: what was measured, what was found, what failed, and what is honestly still unknown.

Peace, people.

Part III · The latest findings

Reports, sealed evidence and the record

The documents behind the claims above. Where a summary on this page compresses a caveat, the sealed files govern.