Research a Decision without Losing Source Boundaries

A decision brief plus claim ledger that shows what each source supports, what remains uncertain, and what would reverse the recommendation.

Strategists, operators, and creators researching changing, contested, or consequential decisions.

• The decision and decision-maker
• Options under consideration
• Freshness date
• Allowed and disallowed source types
• Load-bearing claims
• Constraints and reversal conditions

1. Frame the decision

   Write the action this research must inform, the deadline, and the consequence of error. Interesting information that cannot change the action is secondary.

2. Create the claim ledger first

   List claims required for each option. Classify each as sourced, derived, inferred, guessed, judgment, or unsupported but verifiable.

3. Set the source hierarchy

   For changing facts, prefer current official or primary sources. Record publication date, event date, scope, and the exact claim supported.

4. Compute derived claims

   Show formulas, inputs, units, and assumptions. A sourced input does not automatically make the resulting calculation correct.

5. Preserve conflict

   Record meaningful contradictory evidence and the strongest alternative explanation. Narrow the conclusion when the evidence cannot distinguish them.

6. Draft from the ledger

   Lead with the recommendation at the confidence earned. Attach sources to claims, not paragraphs, and end with unknowns and a reversal condition.

• Primary-source websites or official documents
• Claim Trail
• A calculator or spreadsheet
• A decision brief template
• A dated source ledger

• Human approves the decision frame and source rules
• Human reviews use of private, paid, or licensed material
• Human checks load-bearing calculations
• Decision owner approves the final recommendation

• If a load-bearing claim lacks support, research it, narrow the claim, or mark the recommendation provisional
• If sources conflict, preserve the conflict instead of averaging it away
• If only secondary commentary is available for a changing fact, state that limitation
• If research crosses a private-source boundary, stop before quoting or publishing

• □Every load-bearing claim has a status
• □Every citation directly supports the nearby claim
• □Dates and scopes are visible for changing facts
• □Derived numbers can be independently recomputed
• □The strongest alternative is coherent, not ceremonial
• □Unknowns and reversal conditions could actually change the recommendation

Research this decision with strict source boundaries.

Decision: [action to choose]
Decision-maker: [person or role]
Options: [list]
Deadline/freshness date: [date]
Allowed sources: [rules]
Private or disallowed sources: [rules]
Known constraints: [cost, time, risk, environment]

Before drafting, create a claim ledger. For every load-bearing claim record: status (sourced, derived, inferred, guessed, judgment, or unsupported but verifiable), best primary source, date, exact support, scope limit, contradiction, and consequence if false. Show calculations. Then produce a decision brief with recommendation, tradeoffs, unknowns, confidence capped by the weakest load-bearing claim, and a concrete reversal condition.

Synthetic example

Scenario

A small team must choose between two transcription services for confidential interviews.

Input

Official security pages, current pricing, a synthetic audio benchmark, retention requirements, and the team's monthly volume.

Result

A recommendation separating official retention claims, spreadsheet cost calculations, benchmark observations, and the human judgment that privacy outweighs a small accuracy advantage. A policy change or failed retention test is the reversal condition.