At Deborah's suggestion, I recently turned the CAMS engine onto a society I don't follow closely: South Africa. That distance turned out to be an advantage. It gave me a clean test of whether the model — and different AI platforms — agree.
What follows is a short field report from that experiment — and why it matters for how we read the stress lines in the wider Western system.
Step 1 — Classic Scoring: Gem vs Grok
Eight functional nodes (Executive, Army, Priesthood, Property Owners, Trades, Proletariat, State Memory, Merchants). Four variables per node. Long-run coverage: South Africa 1880–2025.
Two systems run in parallel: a custom Gemini Gem and a custom Grok project with identical instructions. Not wild contradictions — but some differences.
Step 2 — Normalising the Language
Between runs, CAMS moved to universal archetypal node names. A new instruction set was written, and both platforms were re-run.
Executive → Helm · Army → Shield · Priesthood → Lore · Property Owners → Stewards · Trades → Craft · Proletariat → Hands · State Memory → Archive · Merchants → Flow
Same country, same time window, same questions, new lexicon. The scores still didn't numerically snap to one another — but the differences became clearly interpretive, not structural.
Where the AIs Disagreed — and Why That's Reassuring
The platforms disagreed on two things: how "good" or "clean" the democratic transition really was, and whether the Ramaphosa government is genuinely reversing the institutional corrosion of the Zuma era — or merely slowing it.
In a nice twist, when I asked Grok to compare the Gemini and Grok-derived datasets, it volunteered that Gemini had probably done the stronger job on the fine-grained scoring. I didn't prompt that; it was an internal judgement based on consistency and narrative fit.
This is disagreement about emphasis resting on a shared thermodynamic skeleton.
What All Four Systems Agreed On
Consensus across Gem, Grok, DeepSeek, and Claude
- A system that carries high stress for a very long time — first through racial-authoritarian order, then through an extremely demanding transition
- An apartheid regime that maintained Coherence by burning vast Capacity and exporting entropy into the oppressed majority
- A post-1994 system that releases pressure and gains institutional Coherence — but inherits enormous structural Stress in Hands, Craft, and Flow
- A Zuma phase where elite capture erodes Archive and Helm, pushing the system back toward reactive, entropy-heavy behaviour
- A Ramaphosa phase that improves things at the margin — but not enough to restore a high-Ψ (deliberative) regime. South Africa is stabilised, not "fixed"
Why This Matters Beyond South Africa
The thermodynamic dual-mode formalism is robust across tools and perspectives. When multiple AIs describe a country in terms of Coherence, Capacity, Stress, and Abstraction — the numbers differ, but the dynamics do not. The same peaks, troughs, and complex adaptive shifts appear.
| Society | Direction | Key Observation |
|---|---|---|
| USA | ↓ Brittle | High abstraction, falling coherence, rising reactive dominance. Thermodynamically old. |
| China | ↑ Coherent | Strong, metabolically efficient — channelling stress into capacity formation and R&D. |
| Russia | ↑ Resilient | Catastrophic 1990s hit, then high-stress but high-capacity recovery. Post-2022: tightened internal coherence. |
| South Africa | → Stabilised | Honest mirror between worlds. Ψ partial recovery; Φ never fully retreated. |
Abstraction Inversion: When Storytelling Replaces Action
One of the sharpest signatures in the Western data is what CAMS calls abstraction inversion. In a healthy system, abstraction supports deliberation — you spend energy on law, science, and planning so that Helm, Shield, Lore, and Flow can make better use of finite capacity.
In an inverted mode, abstraction becomes a substitute for action. The system pours energy into narratives, branding, messaging, and culture wars — while deferring the painful material adjustments that would actually reduce stress.
The AIs read this in the numbers as: high Abstraction, low or falling Coherence, rising Stress, and a Ψ that cannot grow because it has no spare metabolic headroom.
When I asked Claude for an interesting insight from the data without specific prompting, it went straight to that. See the full thread →
None of this is a moral verdict. It's a reading of metabolism, not virtue. But as the Western system leans ever harder into narrative and away from repair — while non-Western systems respond to pressure by tightening coherence and building capacity — the CAMS lens suggests something very simple and very physical: