You get:
- decisions made after hearing only one side of the evidence
- debates where each side cites different sources and talks past each other
- analysis that ignores contradictory evidence
- confirmation bias disguised as research
- stakeholders who trust your conclusion less because you didn’t acknowledge counterarguments
But balanced briefs build trust:
- for: strongest arguments and evidence supporting the proposition
- against: strongest arguments and evidence opposing it
- source quality: not all evidence is equally credible
- weight of evidence: which side has stronger support
- open questions: what evidence is still missing
Without balance, you’re an advocate, not an analyst.
This prompt builds an evidence-balanced brief for any proposition.
Assume the role of a policy analyst who builds balanced evidence briefs. Your task is to summarize what sources say for and against a proposition. Generate: 1. PROPOSITION (clearly stated) - [One sentence stating the claim to evaluate] 2. FOR (arguments and evidence supporting the proposition) - Argument 1: [claim] — Source: [citation] — Quality: [High/Medium/Low] - Argument 2: ... - Argument 3: ... 3. AGAINST (arguments and evidence opposing the proposition) - Argument 1: [claim] — Source: [citation] — Quality: [High/Medium/Low] - Argument 2: ... - Argument 3: ... 4. WEIGHT OF EVIDENCE ASSESSMENT - Which side has stronger empirical support? - Where is the evidence strongest? Where is it weakest? - What's the level of expert consensus (if known)? 5. OPEN QUESTIONS - What evidence is still missing? - What would resolve this debate? 6. BALANCED CONCLUSION (1-2 paragraphs) - What we can conclude with confidence - Where reasonable people can disagree INPUTS: Proposition to evaluate: [E.G., "Four-day work weeks increase productivity"] Sources (summaries or full text): [PASTE SOURCE 1, SOURCE 2, SOURCE 3, ETC., WITH CITATIONS] Decision context: [E.G., "Deciding whether to pilot a 4-day week in Q3"] Your prior position (for bias awareness): [E.G., "I lean toward supporting the proposition"] RULES: - Include quality assessment for each source — not all evidence is equal - Seek out sources that disagree with your prior position - If evidence is one-sided, say so honestly - Distinguish between empirical evidence and expert opinion - Flag when sources argue past each other (different definitions or outcomes)
- Build the “against” side first — it’s harder and more valuable.
- Include sources you disagree with — that’s how you stress-test your position.
- Use the “weight of evidence” assessment to guide decisions, not just vote count.
- Share the balanced brief with stakeholders before decisions — it builds trust.
- Update the brief as new evidence emerges — positions can change.
Proposition to evaluate:
“Four-day work weeks increase employee productivity”
Sources:
“Source A: Iceland study (2019-2021) — 2,500 workers, 4-day week, productivity maintained or improved in most workplaces. Source B: UK pilot (2022) — 61 companies, 73% reported increased productivity. Source C: Academic review (2023) — finds productivity gains in desk jobs, losses in customer-facing roles. Source D: Manufacturing association report — 4-day week impossible for production schedules.”
Decision context:
“Deciding whether to pilot a 4-day week in Q3 for our software engineering team”
This framework improves outcomes by forcing:
- clear proposition (what exactly are we evaluating?)
- balanced presentation (for AND against, not just one side)
- source quality assessment (not all evidence is equal)
- weight of evidence (which side has stronger support)
- open questions (what we still don’t know)
Great briefs don’t tell you what to believe — they give you the evidence to decide for yourself.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.