Methodology

Every story starts as a signal: a dated, attributable observation from a named publisher, regulator, dataset or primary document. Signals without a source and a timestamp are not used.

Evidence standard

A claim needs a source that a reader can check. Where two sources disagree, the disagreement is reported rather than resolved silently. Where a number is estimated, the estimate is labelled as one.

The seven lenses

• Biological: What does the body notice, remember, desire or avoid?
• Cultural: What shared meaning, identity or social code is operating?
• Analytical: What does the evidence show, and where is the measurement lying?
• Journalistic: What happened, who benefits and what is missing?
• Pedagogic: What must the audience understand before it can decide?
• Performative: How does delivery, ritual, spectacle or charisma change belief?
• Endocrinological: How do stress, reward, arousal or safety shape action?

Reporting versus commentary

Reported facts appear in body text. Interpretation appears in clearly marked commentary blocks and in the bottom line. Readers should never have to guess which is which.

How a reading is produced

Readings are assembled automatically. Collectors gather public documents, a language model groups them and writes the reading, and the result is published without a person editing it first. That means a reading can misread an organisation, attach evidence to the wrong entity, or state an implication the evidence does not carry. Treat it as a research instrument, not a verified report, and follow the evidence links before acting on anything commercially significant.

Reading the confidence score

Every insight carries a confidence score, and it is your barometer of accuracy. It summarises how strong the underlying evidence is: the quality and independence of the sources, how many of them agree, how recent they are, how certain the system is that it identified the right organisation, and whether anything contradicts the reading. A low score does not mean the observation is false — it means the evidence behind it is thin, narrow or recent, and the claim should be treated as a lead rather than a finding.

• 0–34% Insufficient: weak or single-source material. Not a basis for a decision.
• 35–49% Emerging: a hypothesis worth watching.
• 50–64% Directional: the direction is probably right; the detail may not be.
• 65–79% Substantiated: corroborated by independent sources.
• 80–100% Strongly substantiated: multiple independent, high-quality sources agree.

Each insight page opens its own calculation, so you can see which components raised or reduced the score. Demonstration content is labelled as such everywhere it appears.