Neural Nations
One representation for every society, including our own.
The starting claim is that a human society belongs to the broad family of bonded, surplus-dependent adaptive systems. A nation is not literally an enlarged chimpanzee troop, yet beneath the institutional differences lies an older problem. Any social unit must coordinate action, manage conflict, distribute resources, preserve knowledge and respond to threat. The comparison does not reduce civilisation to zoology; it identifies the functional domain from which civilisation emerged. Its meta-animal is functional, not racial; diagnostic, not normative. A score is a state, not a grade. A trajectory is a history, not a destiny.
Independent research, not formally peer-reviewed. Every dataset, formula and scoring protocol is published openly. Scrutiny is not a challenge to this project; it is the point.
Why "Neural Nations"?
The name is technically misleading. Nations are only one scale at which this coordination anatomy appears. The same patterns show up in empires, civilisations, corporations, and organised collectives of any kind. "Neural Nations" is a misnomer — but it's poetically useful. It suggests something deeper than politics: the nervous system of organised human life.
A Note on Method
CAMS compares societies through eight functions of human coordination. It tests whether representing those functions consistently reveals patterns obscured by separate national narratives. Whether its eight functions, scoring procedures and derived measures recover useful structure must be established through reproducibility, independent evidence, comparison with alternative models and prospective tests (the August 2026 register and the 2027–2028 register). No society receives an exemption from comparison; no part of the instrument receives an exemption from revision.
Where Things Stand (November 2026)
The current statement of the framework is The Natural History of Civilisations. It brings together the essay, the evidence behind it and links to every result below. In brief:
- Era recognition. In the November 2026 replication round, AI systems were given only an eight-number portrait of each era. On the 56-era core panel (Russia, Australia and the USA), Claude and Gemini each matched every portrait to its era from a closed list, and Kimi matched 52. Recognition weakened on the extension panel and on the sparse Denmark control. The pre-registered null models were never returned. This is a promising demonstration, not an external validation. Report · repeat it yourself.
- Sealed predictions, scored. In ADVERSARY-02, 15 predictions for China and seven firms over 2021–2024 were sealed before any data was looked at. Six passed and nine failed, and the built-in kill switch was not triggered. Results.
- Prospective test lodged. 32 probability-weighted claims about the USA, China and Australia in 2027–2028 are lodged at Zenodo, to be scored when those years close. Register.
- Pathologies across scales. On JUNO v3 (42 societies), with thresholds frozen before counting, four named signatures concentrate in their windows beyond chance. Twelve of eighteen named windows hold. Interim results.
Failures are part of the record: see Validation & Limits and the lab diary.
Where to Start
Also worth reading
Kari McKern — researcher, former public servant, IT specialist, and geopolitical analyst based in Sydney. CAMS began as a question about modern geopolitics and became a framework for civilisational dynamics. The data are open. The model is documented. The work continues.
[email protected] ·
"Patrolling the event horizon between the known and the unknown."