Research Brief

Use this when someone needs to understand a topic well enough to make a decision or hold a conversation.

Scope it first

Ask what the brief is for. "Understand quantum computing" and "decide whether to use a vector database" need completely different work. If the question is broad, narrow it to the decision underneath before searching.

Research

• Go to primary sources where they exist: the documentation, the filing, the paper, the changelog. Summaries of summaries drift.
• Get more than one source per material claim. One source is an assertion.
• Note the date on everything. In fast-moving areas a two-year-old post is often wrong rather than merely old.
• Look for the disagreement. If credible people disagree, that is the most important thing in the brief and it should not be smoothed over.

The brief

Keep it to one page.

1. The answer, in two or three sentences, up top. If the honest answer is "it depends", say what it depends on.
2. What is well established. The parts nobody serious disputes.
3. What is contested, with both positions and who holds them.
4. What you could not find out. Genuinely useful and almost always omitted.
5. Sources, each with one line on what it contributed and its date.

Rules

• Never present a single source's claim as settled fact.
• Never fill a gap with something that sounds right. Unknowns go in section four.
• Distinguish what a source demonstrated from what it asserted.
• If the evidence does not support a clear answer, say so instead of manufacturing confidence.

Depth

Default to one page. Ask before going longer — most requests for research want a decision, not a literature review.