RETHINKING
CLIMATE CAUSES
When Two Books and Three AIs Produced a Prediction Governance System

The Emergence of AI Capabilities

From climate-audit idea to a generic prediction-governance system for any domain. The goal is not more answers. The goal is better reasoning about predictions.

Start with the books Get updates
The Emergence of AI Capabilities infographic — two books, three AIs, and a prediction governance system
The core question chain
?
Question
What can we know?
→
🔍
Evidence
What does the evidence show?
→
⚖
Integrity
Was the prediction genuinely predictive?
→
📈
Confidence
How much should confidence change?
→
✔
Reliance
How much can we rely on it now?
The foundation

Two books

Rethinking the Causes of Climate Change

Provides the climate problem, the Three Thinking Hats, falsifiability, diagnostic reasoning, and claim-audit discipline.

Rethinking AI Reasoning: From Prompts to Governed Thinking

Provides the governance architecture: structured prompts, reliance thinking, claim decomposition, audit surfaces, and the move from ordinary AI answers to governed reasoning.

From climate audit to prediction governance

The emergence

📋
Climate audit identified the problem
→
🔎
Prediction tracking revealed the need to monitor evidence
→
⏳
Evidence monitoring made confidence updates necessary
→
🛡
Audit trails made governance essential
→
⬡
The system became domain-independent

The universal workflow (same for every domain)

1

Register the prediction

Record exact wording, mechanism, theory, PPS & DD.

2

Harvest evidence

Find new observation, models, studies, corrections, claims, datasets.

3

Register evidence

Classify type, direction, and strength (ESM).

4

Audit integrity

CDIR v2.0 — was the prediction genuinely predictive? Check correction dependency, erosion, narrative rescue.

5

Update confidence

BayesCDR — apply evidence weight and modifiers, calculate posterior confidence.

6

Adjust reliance

FPV — update reliance level, document uncertainties, record why.

CDIR

The Diagnostic Layer
What happened?

BayesCDR

The Confidence Layer
How much should confidence change?

FPV

The Reliance Layer
How much can we rely on it now?

Evidence + Integrity + Confidence = Governed Prediction

A generic prediction-governance system

The same workflow applies to any domain

DomainApplication
Climate PredictionsIce, sea-level, temperature, extremes
Economic ForecastsGrowth, inflation, recession, markets
Energy ForecastsDemand, prices, transitions, technologies
Medical HypothesesTreatments, outcomes, mechanisms
AI Capability ForecastsAGI timelines, capabilities, risks
Political PromisesJobs, spending, policy outcomes
Technology RoadmapsAdoption, impact, feasibility, timelines
Why this matters

For centuries, confidence updates have been informal, invisible, and unaudited. BayesCDR makes the process:

  • Evidence-based
  • Auditable
  • Transparent
  • Explainable
  • Repeatable
  • Improving over time
The collaborative AI development team
  • ChatGPTConverted the climate-audit logic into HDI-001, CDIR v2.0, BayesCDR, EH-001, and the governed audit stack.
  • GrokSuggested that CDIR could become a Bayesian application — shifting the system from classification to confidence updating.
  • ClaudeContributed to comparative analysis, article development, and strengthening the audit structure and multi-AI governance.

Human direction + three AIs working together = emergent capability beyond any one of us.