How We Analyze

High Confidence

Strong evidence from multiple independent, reliable sources. Key judgments are well-supported and alternative explanations have been considered and addressed.

Moderate Confidence

Reasonable evidence but with some gaps. Sources generally agree but may lack diversity or independence. Alternative explanations exist and cannot be fully ruled out.

Low Confidence

Limited or conflicting evidence. Significant uncertainty remains. The assessment represents the best available judgment but could change substantially with new information.

Grade A, Completely Reliable

Wire services (Reuters, AP), official government statistics, peer-reviewed journals, and central bank publications.

Grade B, Usually Reliable

Major broadsheet newspapers, established think tanks (Brookings, CSIS), international organizations (World Bank, IMF), and recognized industry analysts.

Grade C, Fairly Reliable

Regional quality news outlets, academic institutions, NGOs with domain expertise, and specialized industry publications.

Grade D, Not Usually Reliable

Unknown commercial domains without established track records. These sources are flagged and given lower weight in analysis.

Grade E, Unreliable

Social media platforms, personal blogs, and unverified user-generated content. Included only when no better source exists, and always flagged.

What they are

Every analysis applies structured methods drawn from professional intelligence practice, domain-specific framing, key questions, and structured analytic techniques appropriate to the topic.

Cognitive bias awareness

Analytical methods are designed to surface competing explanations and flag where common reasoning patterns, confirmation bias, anchoring, availability bias, may distort conclusions. This is built into how analyses are structured, not a post-hoc check.

Why it matters

Unstructured analysis is prone to systematic errors that go unnoticed. Applying the same disciplined methods every time reduces the chance that a conclusion is an artifact of how the question was framed.

What it means

The number of independent sources consulted during analysis. Higher counts indicate broader evidence collection. Deep assessments draw from a substantial source base across multiple source types.

Why it matters

More sources reduce the risk of relying on a single perspective and increase the robustness of conclusions. Source diversity (government, academic, industry, media) matters as much as quantity.

Assessment

Deep-dive analysis with extensive source coverage, structured analytical techniques across relevant intelligence domains, and calibrated confidence on every key finding.

Intelligence Brief

Timely analysis of trending developments with structured analytical techniques and source grading. Published regularly.

Flash

Quick-turn analysis of breaking developments. Concise and focused, designed for rapid consumption.

What they are

The most important analytical judgments from the analysis, each tagged with a confidence level. Key findings represent conclusions, not just facts, they include the analyst's assessment of what the evidence means.

How to read them

Each finding pairs a confidence indicator with a substantive judgment. "High Confidence: X is happening" means strong evidence supports that conclusion. "Moderate Confidence: Y is likely" means the evidence points that direction but with notable uncertainty.

The same event has tactically-minor meaning to one actor and strategically-existential meaning to another. Forecasts from a single actor-scale produce systematically miscalibrated predictions.

What an actor can do ≠ what they will do. Analyses anchored to capability over-predict action; analyses anchored to intent over-predict restraint. The truth lives in the gap.

When the participants reading the analysis change the outcome being analyzed — markets react to the recession forecast; the electorate shifts on reading the poll — standard forecasting fails on reflexive systems.

Accumulated state (total debt, installed capacity, ice mass) versus rate of change (deficit, generation, melt rate) produce different forecasts. Conflating them misses inflection points.

What looks like "one actor" — a coalition, alliance, party, cartel — is often many actors who may defect under stress. Treating coalitions as unitary actors misses the defection cascade.

Short-term tactical wins create long-term strategic losses, and inversely. Forecasts pinned to one horizon miss the cross-horizon trade-off where the asymmetry lives.

Insider sources have access but bias. Outsider sources have distance but lose signal. The truth requires triangulating both — neither pool alone closes the picture.

What gets reported ≠ what happens. Survivorship in startups, media-coverage skew, and intelligence-channel selection produce a sampled view of reality that systematically distorts forecasts.

Causal claims ("X caused Y") cannot be assessed without "what would have happened without X." The counterfactual is the only access route to causation; without it, analysis is correlation dressed as causation.

The analyst's historical track record per domain is the only honest bound on confidence in future forecasts. Without per-domain calibration, all confidence statements are aspirational.