Start with the record
Primary documents, regulatory filings, government data, and peer-reviewed research come first. Local journalism, advocates, labor, and community groups matter when they surface facts absent from official disclosures.
About the research
Trust is a method,
not a badge.
BrooksFlow is designed to say “we know,” “we estimate,” “experts disagree,” and “we do not know yet” with equal clarity.
Label what kind of knowledge it is
Observed, estimated, forecast, announced, disputed, early evidence, and local are not interchangeable. The label travels with the claim everywhere it appears.
Keep geography attached
A national average is not a local answer. Case studies retain their utility territory, watershed, state, market, and policy context.
Show the uncertainty
Forecasts appear as ranges when credible estimates diverge. Assumptions and caveats are part of the visual, not hidden in an appendix.
Review for freshness
Every claim carries a last-reviewed date. New evidence can update one structured record and propagate through every page that uses it.
Correct in public
Substantive corrections should identify what changed, why, and when. Suggestions are welcomed; unsupported edits are not silently accepted.
Evidence language
Tell people what kind of number it is.
Three labels travel together: what produced the evidence, its time status, and the form of the result.
- Observed
- Measured historical result.
- Forecast
- A modeled future range, never an accomplished fact.
- Sponsor-reported
- A company's figure; useful, but not independently verified.
- Commitment
- A promise or obligation, not a measured result.
- Required
- A binding law, tariff, permit, or order—kept separate from guidance.
- Randomized trial
- A controlled comparison designed to estimate an intervention's effect.
- Observational
- A measured relationship that does not by itself prove causation.
- Unknown
- Not disclosed or not verified. Never converted to zero.
Every numeric claim
Evidence has coordinates.
- Time
- Historical · current · projected · conditional
- Result form
- Absolute · relative · percentage-point · forecast band · modeled scenario
- Relationship
- Independent · primary · sponsor-affiliated · underlying study group
- Place
- Nation · state · utility territory · county · facility · watershed
- Limits
- Assumptions · competing evidence · local interpretability · missing disclosure
- Vintage
- Publication date · as-of date · last review · superseded claim