The Scale Invariance Experiment

Blind Inference from CAMS Time-Series — §5.1 Supplementary Data

If the Complex Adaptive Model State captures something real about how civilisations are structured, then an AI system given only anonymous numerical time-series — no names, no geography, no context — should still be able to recover the structural archetype of each entity. This is the scale invariance test: the pattern speaks for itself.

Protocol. Five societies were selected for structural diversity: one post-war reconstruction arc, one developmental state, one long imperial–republican transition, one settler-colonial democracy, and one city-state. Their identities are blinded. Each dataset has had its entity label stripped, its year axis jittered by ±1, and Gaussian noise (σ = 0.3) added to the four raw metrics. Node Values and Bond Strengths are recomputed from the noisy metrics. The challenge: identify the structural archetype — and if possible, the entity — from the numbers alone.

Blinding note. The mapping between file numbers and societies is held separately and not published alongside these files. If you are participating in the inference challenge, do not look at the generation script until after you have submitted your identification.

Download Blinded Datasets

Five CSV files · 1900 onwards · 9 columns · identity replaced with MarkerXXX

All files also available in the GitHub repository under blinded/.

Column Schema

Column	Description	Range
Society	Blinded entity label — MarkerXXX, not the real name	Marker001–005
Year	Calendar year, jittered ±1 from original	integer
Node	One of eight institutional nodes	Archive · Craft · Flow · Hands · Helm · Lore · Shield · Stewards
Coherence	Internal alignment and cohesion of the node	1–10
Capacity	Material and organisational resources	1–10
Stress	Load and entropy pressure on the node	1–10
Abstraction	Symbolic and cognitive complexity	1–10
Node Value	Coherence + Capacity + (Abstraction/2) − Stress	≈ −7 to +24
Bond Strength	Mean pairwise coupling for this node across all 7 partners	≈ 0.5–4.5

Node Value = C + K + A/2 − S Bond Strength_i = mean_j≠i √( max(V_i+8, 0) · max(V_j+8, 0) ) / 8

Each row represents one institutional node for one year — 8 rows per year-step. Gaussian noise (σ = 0.3) was applied to the four raw metrics; Node Value and Bond Strength were recomputed from the noisy values.

Key Finding

Analysis of five blinded societies spanning 70–125 years reveals strong evidence of scale invariance: the same organisational principles, node relationships, and crisis signatures repeat across all societies regardless of their specific historical context or scale. This suggests CAMS captures something fundamental about how human societies must organise themselves to function.

→ CAMS Scale Invariance: What It Reveals About Universal Social Dynamics

Experiment Report

The full write-up of the scale invariance experiment — methodology, blind inference results, recovery accuracy, and structural conclusions — is available as a PDF.

↓ CAMS Scale Invariance Report 2026 (PDF) ↓ CAMS Scale Invariance Report — earlier version (PDF)

The Inference Challenge

Feed one or more of these files to any capable LLM or statistical classifier. Provide only the CSV — no hints. Ask it to characterise the structural archetype of the entity: its dominant institutional regime type, probable historical period, trajectory shape, and if the signal is strong enough, a candidate identification.

A valid recovery requires: (1) correct quadrant assignment in phase space, (2) identification of at least two major structural transitions visible in the data, and (3) a structural archetype label consistent with the CAMS typology. Naming the actual entity is the strong claim — the experiment does not require it.

Submit findings to [email protected] with subject line Scale Invariance Submission. Results will be published in the research diary once the blind is broken.