Kari Freyr McKern — Complex Adaptive Humans CAMS Formulation 3.0 — March 2026
This analysis treats China as a complex adaptive system and measures its structural health across 125 years (1900–2025) using five independently generated datasets. Rather than relying on any single historical narrative, the analysis computes standardised diagnostics from raw institutional data and tests them against pre-registered falsification criteria. The framework — the Complex Adaptive Model of Societies (CAMS) — models every society through eight functional institutional pillars, each measured across four dimensions: internal alignment (Coherence), resource effectiveness (Capacity), material strain (Stress), and symbolic complexity (Abstraction). The interaction between these pillars, measured through coupling strength, determines whether a society can coordinate itself or is approaching a phase transition — what conventional language calls a crisis.
The central finding is that China's structural trajectory over 125 years tells a story that neither triumphalist nor declinist narratives capture. The data reveals a system that has achieved remarkable material recovery since 1978, now operating near its historical peak in composite health — but carrying three persistent structural vulnerabilities that predate the current regime and have survived every political transformation of the twentieth century.
The analysis draws on five independently generated datasets, each representing a distinct AI assessor's reconstruction of China's institutional history. Two long-run datasets (Gemini-Nov and GPT-Jan27) cover 1900–2025; a third long-run dataset (Gemini-1900-2025) covers the same period with a different scoring convention; one early-period specialist dataset (Gemini3-Jan28) covers 1880–1935; and one recent-period specialist (Gemini-Feb28) covers 2010–2025.
The ensemble approach is methodologically critical. Where independent assessors agree, we have evidence of structural reality. Where they disagree, we identify scoring artefacts or genuine measurement uncertainty. The five datasets collectively provide 443 annual time-slices for analysis.
Concordance testing reveals that the two primary long-run datasets (Gemini-Nov and GPT-Jan27) agree strongly on the major variables: the correlation on mean pillar health is ρ = 0.675, on stress is ρ = 0.706, and on coupling strength is ρ = 0.804. All of these comfortably exceed the ontological safety threshold of ρ > 0.3 specified in the framework's falsification protocol. The third long-run dataset uses an inverted stress polarity and a different bond-strength scaling convention, which explains its lower concordance with the other two. Once this is accounted for, no dataset fundamentally contradicts the others.
This level of agreement — achieved across different AI systems using different prompting contexts and producing assessments months apart — is strong evidence that the structural dynamics described here are not artefacts of any single assessor's biases.
CAMS specifies explicit conditions under which the framework would require revision. Two of these were tested directly on the China ensemble.
The framework predicts that stress and capacity must be negatively correlated across all nodes — a society's institutional pillars cannot simultaneously experience high strain and high effectiveness. Four of five datasets confirm this at high significance, with correlations ranging from ρ = −0.53 to ρ = −0.80. The fifth dataset shows an apparent violation (ρ = +0.71), but inspection reveals this is a coding artefact: that dataset stores stress as negative values (more negative = more stressed), which inverts the correlation sign. Once polarity is corrected, all five datasets confirm the thermodynamic constraint. The universality of stress–capacity anti-correlation holds for China.
The framework predicts that year-to-year changes in system health are dominated by changes in how well institutional pillars coordinate with each other — not by changes in any single pillar's condition. This was tested by correlating annual changes in coupling strength against annual changes in mean pillar health. The results are among the strongest in the entire CAMS archive: correlations range from ρ = 0.897 to ρ = 0.982 across all five datasets. This means that coupling between pillars explains 80–96% of the variance in China's year-to-year health changes. The implication is direct: what matters most for China's stability is not the condition of any individual institution, but how well institutions work together.
The ensemble data divides naturally into nine historical phases, each with a distinctive structural profile. What follows is not a political narrative but a measurement of how the system's internal coordination shifted across each period.
Mean pillar health of 6.8 (on a theoretical scale where values below ~5 indicate systemic distress and above ~12 indicate strong health). Negative collective affect — the system was experiencing net strain across most pillars. All assessors independently flag executive decoupling: the governing apparatus had structurally disconnected from the institutions it was meant to coordinate. The Qing dynasty did not fall because of any single policy failure; it fell because the coordination graph had already fragmented.
The worst pre-Communist period by every metric. Mean pillar health drops to 5.3, the criticality index (measuring institutional shear — how fast different pillars are pulling apart) reaches its highest value at 0.147, and collective affect falls to −1.3. This is the structural signature of a system with no effective central coordination: each pillar is evolving at its own rate, in its own direction, with no coupling mechanism to hold them together.
A measurable recovery. Pillar health climbs to 8.0, the criticality index drops, and collective affect returns to marginally positive territory. The Nationalist government partially re-synchronised the institutional fabric — a genuine structural achievement that is often underestimated in revolutionary historiography.
The Japanese invasion and subsequent civil war drive pillar health back to 6.2 with negative affect. However, coupling strength actually rises during this period (to 4.48), and the criticality index drops sharply (to 0.085). This is the structural signature of wartime mobilisation: institutions align tightly around survival, even as material conditions deteriorate. The system is stressed but synchronised — a configuration that historically precedes either rapid recovery or total collapse, depending on what follows.
The new regime lifts pillar health to 7.5 and reduces mean stress to its lowest level so far (4.6). Collective affect turns marginally positive. But the period also carries the highest dissociation index yet (Σ = 6.0), indicating that some pillars — particularly the knowledge-producing institutions — are sustaining high symbolic complexity under significant material strain. The Great Leap Forward (1958–1962) creates a sharp spike in executive decoupling detected across multiple assessors.
The absolute nadir. Mean pillar health collapses to 3.9 — the lowest in the entire 125-year record. The dissociation index spikes to 8.0, its historical maximum. Collective affect falls to −1.7. Institutional cognition — the product of abstraction and alignment — crashes to 18.4, roughly half its pre-revolutionary level. Every assessor independently flags simultaneous executive decoupling, military-security dominance, and either market collapse or narrative–productive decoupling. This is as close to thermodynamic death as modern China has come: a system where adaptive function has nearly ceased, stress is uncontrolled, and the coordination graph has been deliberately disrupted from within.
Recovery begins. Pillar health climbs to 7.2, and coupling strength rises to 4.45 — above the pre-revolutionary baseline. Collective affect returns to zero. The military-security dominance signature persists (the security apparatus remains disproportionately strong relative to other pillars), but material conditions improve. The structural story of the Deng reforms is not primarily one of ideological liberalisation; it is one of institutional re-coupling.
China's structural golden age. Pillar health reaches 10.4, coupling strength hits its all-time high (5.48), and the criticality index drops to its all-time low (0.068). The dissociation index is nearly zero. Collective affect is positive (1.6) — the only period in which a substantial majority of institutional pillars are experiencing genuine surplus. The soul reading shifts from transitional to what the framework terms "late hypertrophy" — a system whose cognitive sophistication (37.2) is growing faster than its affective margin. This is not a warning sign in itself, but it identifies the trajectory that matters for what comes next.
Mean pillar health rises further to 11.3, the highest in the record. Institutional cognition reaches 43.6 — also a historical peak. But three metrics have moved in the wrong direction relative to the growth era: coupling strength has declined from 5.48 to 4.73, the criticality index has risen from 0.068 to 0.085, and the dissociation index has climbed from near-zero to 6.0. Collective affect remains narrowly positive (1.8), but the margin is thin.
The structural reading is precise: China is more sophisticated but less well-coupled than it was fifteen years ago. The system has become cognitively richer — its institutional pillars are more complex, more abstract, more capable of symbolic processing — while the bonds between them have frayed. This is the signature the framework calls Late Hypertrophy: high abstraction sitting on a narrowing coordination margin.
The ensemble reveals three structural features that appear across all datasets, all assessors, and all political regimes. They are not artefacts of any particular era's politics; they are properties of the system itself.
The Stewards pillar — representing property rights, civil governance, and wealth management — is the weakest institutional pillar in every dataset, with an ensemble-averaged health score of 3.99. This is roughly half the ensemble average (8.07 for Helm, 9.08 for Shield). Stewards also carries the most negative collective affect of any pillar (−1.52), meaning it consistently operates under net strain. This deficit is not a feature of communist governance alone; it appears in the late Qing, the warlord period, the Nationalist era, and every subsequent regime. It is a 125-year structural constant.
The Shield pillar (military, security, and coercive apparatus) is the strongest pillar in the ensemble (V = 9.08) with the most positive collective affect (+1.07). Multiple assessors flag a Praetorian signature — a condition in which the security apparatus becomes disproportionately strong relative to the productive base — across large portions of the twentieth century. The Praetorian signature intensifies after 1949, briefly abates during the 2000s growth period, and returns in the 2015–2025 period. This does not mean China is a military dictatorship; it means that the security pillar consistently absorbs a disproportionate share of the system's coordination capacity.
The gap between what the narrative apparatus asserts (Lore pillar: culture, ideology, education, media) and what the productive base delivers (Hands pillar: labour force, manufacturing, agriculture) is a near-permanent feature of the Chinese system. The Mythic–Material Decoupling signature — flagged when the gap between Lore and Hands exceeds 8 points and they carry opposite signs — appears almost continuously from 1910 to present in the Gemini-Nov dataset. Other assessors flag it less persistently but still identify it as a recurring structural feature. Lore carries the highest dissociation index of any pillar (Σ = 0.65), meaning it is the pillar most consistently operating under high symbolic complexity concurrent with material strain.
This is not a uniquely Chinese phenomenon — narrative–material gaps appear in every society in the CAMS archive. But the persistence and depth of the gap in China is among the most pronounced in the dataset.
The CAMS phase transition theorem classifies societies by how their coupling responds when the criticality index exceeds a threshold — when institutional shear overwhelms bonding capacity. The four possible responses are: re-synchronisation (coupling recovers through coercive or institutional repair), oscillation (coupling alternates between recovery and degradation), fracture (coupling degrades irreversibly), and buffering (the system absorbs shocks before shear becomes critical).
The five datasets disagree on China's attractor class, which is itself revealing. One places China in the re-synchronisation class, one in oscillation, and two in fracture. The ensemble reading is that China occupies the boundary between oscillation and re-synchronisation — a system that has repeatedly recovered from high-criticality episodes through forceful institutional re-coordination, but whose coupling margin is narrowing with each cycle. The current period shows higher baseline criticality (κ = 0.085) than the growth era (κ = 0.068), suggesting the buffering capacity that characterised the 1993–2007 period has partially eroded.
The framework predicts that slow-evolving pillars (governing, cultural, archival, and civil-society institutions) should have longer memory half-lives than fast-evolving pillars (military, productive, commercial, and craft institutions). This prediction holds across all three long-run datasets for China. Archive (state memory, bureaucracy) consistently shows the longest memory — 4 to 9 years depending on the assessor — while Shield (military) shows the shortest (1.5 to 4.7 years). This confirms that the timescale partition — the division of institutional pillars into slow and fast loops — is structurally real for China.
As of 2020–2025, the ensemble provides the following structural snapshot:
Mean pillar health (V̅): 10.89 — near the all-time peak of 12.81 (reached in 2006). China is a historically healthy system by its own standards.
Coupling strength (B): 4.62 — above the long-run average (~4.0) but below the 2006 peak of 5.48. Institutional coordination has declined from its high-water mark.
Criticality index (κ): 0.10 — moderate. Not in crisis, but higher than the 0.068 floor of the growth era. The gap between institutional shear and bonding capacity has widened.
Cognition: 44.5 — the all-time peak. China's institutions are more symbolically complex and cognitively sophisticated than at any point in the record.
Collective affect: +1.3 — narrowly positive. The system has a small margin of capacity over stress, but significantly less than the +1.6 of the growth era.
Dissociation index (Σ): 6.0 — elevated. Some pillars are sustaining high abstraction under significant material strain, the configuration that the framework identifies as a precursor to late-phase coordination failure.
Several common analytical claims about China do not survive contact with the structural evidence.
"China is uniquely fragile because of its authoritarian system." The data shows the opposite of uniqueness. China's structural profile — strong security apparatus, weak civil society, narrative–material gap — clusters with Russia, Iran, and Japan in the broader CAMS archive. These are structurally similar societies despite radically different regime types. The crisis signatures in democracies and autocracies are identical under the framework's diagnostics. Regime type does not confer structural immunity, and it does not create structural uniqueness.
"China's rise is historically unprecedented." In structural terms, the 1978–2007 trajectory is a re-synchronisation from a deep trough — a pattern that appears in multiple societies after catastrophic coordination failure. What is distinctive is not the rise itself but its magnitude and speed: the coupling recovery from 3.43 (Cultural Revolution) to 5.48 (growth peak) is among the largest in the archive.
"The current slowdown signals imminent collapse." The data does not support this. A pillar health of 10.89 with positive collective affect and moderate criticality is not a crisis configuration. What the data does show is a narrowing margin: coupling is declining, the dissociation index is rising, and the criticality index has increased. This is a system moving toward a coordination boundary, not a system at a coordination boundary. The difference matters.
"Xi's consolidation has strengthened institutions." Partially supported, partially contradicted. Pillar health and cognition have continued to rise, suggesting institutional capacity has grown. But coupling strength has declined and the Praetorian signature has returned, suggesting the method of consolidation — centralisation of coordination through the security pillar rather than through civil-society and market mechanisms — carries structural costs that do not appear in aggregate health statistics.
The ensemble analysis suggests three structural dynamics that will shape China's trajectory:
First, the Stewards deficit is the binding constraint. At a health score of 3.99 — roughly half the system average — civil society, property rights, and independent wealth management constitute the system's weakest link. Every other major economy in the CAMS archive has achieved sustained high health only when Stewards was comparably strong to other pillars. China has not yet done this, despite 45 years of reform.
Second, the coupling–cognition divergence is the emerging risk. A system that grows cognitively richer while becoming less well-coupled is a system whose institutional pillars are becoming more capable individually but less able to work together. This is the thermodynamic precondition for what the framework calls non-agentic crisis — a coordination failure that does not require any single institution to fail, only that they fail to synchronise.
Third, the Praetorian return is a structural regression. The brief period during which China's security pillar was not disproportionately dominant (roughly 2000–2014) coincides precisely with the period of highest coupling and lowest criticality. The return of Praetorian signatures after 2015 suggests the system is reverting to a coordination mode — security-mediated synchronisation — that works in the short term but historically narrows the coupling base.
None of these dynamics implies imminent crisis. They describe a trajectory — a direction of structural change — not a destination. The ensemble data shows that China has demonstrated extraordinary capacity for institutional re-synchronisation in the past. Whether that capacity remains available under current conditions is the structural question that matters.
| Label | Source | Period | Annual Slices |
|---|---|---|---|
| Gemini-Nov | Google Gemini | 1900–2025 | 125 |
| GPT-Jan27 | OpenAI GPT | 1900–2026 | 127 |
| Gemini3-Jan28 | Google Gemini | 1880–1935 | 51 |
| Gemini-Feb28 | Google Gemini | 2010–2025 | 16 |
| Gemini-1900-2025 | Google Gemini | 1900–2025 | 124 |
| Criterion | Result | Detail |
|---|---|---|
| FC1: Stress–Capacity | ✓ HOLDS (4/5 direct; 5th polarity artefact) | ρ ranges: −0.53 to −0.80 |
| FC3: Coupling Primacy | ✓ HOLDS (5/5) | ρ ranges: 0.897 to 0.982 |
| Pillar | Health (V) | Stress (S) | Affect (K−S) | Cognition (A×C) | Loop |
|---|---|---|---|---|---|
| Helm (governance) | 8.07 | 5.99 | −0.28 | 35.4 | Slow |
| Shield (security) | 9.08 | 5.22 | +1.07 | 31.1 | Fast |
| Lore (narrative) | 8.22 | 4.82 | +0.17 | 36.2 | Slow |
| Archive (memory) | 8.93 | 5.60 | −0.15 | 41.0 | Slow |
| Hands (labour) | 6.24 | 5.70 | −0.22 | 15.9 | Fast |
| Craft (skill) | 8.61 | 4.85 | +0.96 | 29.8 | Fast |
| Flow (commerce) | 7.63 | 5.24 | +0.28 | 27.2 | Fast |
| Stewards (civil) | 3.99 | 5.77 | −1.52 | 16.8 | Slow |
| Pillar | Half-life η (years) | Loop |
|---|---|---|
| Archive | 6.7 | Slow |
| Stewards | 16.5 | Slow |
| Lore | 6.1 | Slow |
| Helm | 4.9 | Slow |
| Flow | 10.2 | Fast |
| Craft | 10.4 | Fast |
| Hands | 4.4 | Fast |
| Shield | 3.4 | Fast |
CAMS Version 3.0. Formulation: Kari Freyr McKern, Complex Adaptive Humans, March 2026. The framework treats societies as thermodynamic systems of eight coupled institutional pillars and models crises as coordination phase transitions governed by physical law. Full specification available on request.
CAMS v3.0 — Complex Adaptive Humans — March 2026