| name | ratn - Rationale Construction |
| description | Builds a clear, compelling rationale for any decision or recommendation. Organizes evidence by strength, addresses counterarguments, and outputs a persuasive but honest case. |
| output | {"format":"prose"} |
Rationale Construction
Input: $ARGUMENTS
Step 1: State the Conclusion
Lead with the answer. A rationale works backwards from the conclusion.
CONCLUSION: [The decision, recommendation, or position being justified]
AUDIENCE: [Who needs to be convinced?]
STAKES: [What happens if the audience is not convinced?]
Step 2: List the Supporting Evidence
Gather all evidence that supports the conclusion.
SUPPORTING EVIDENCE:
1. [Evidence 1]
Type: [DATA / EXPERIENCE / LOGIC / AUTHORITY / ANALOGY]
Strength: [STRONG / MODERATE / WEAK]
Source: [Where this evidence comes from]
2. [Evidence 2]
Type: [DATA / EXPERIENCE / LOGIC / AUTHORITY / ANALOGY]
Strength: [STRONG / MODERATE / WEAK]
Source: [Where this evidence comes from]
3. [Evidence 3]
Type: [DATA / EXPERIENCE / LOGIC / AUTHORITY / ANALOGY]
Strength: [STRONG / MODERATE / WEAK]
Source: [Where this evidence comes from]
...
EVIDENCE TYPES PRESENT: [Which types do you have? Which are missing?]
Step 3: Organize by Strength
Arrange evidence from strongest to weakest. Lead with your best.
EVIDENCE HIERARCHY:
1. [Strongest evidence] — this alone almost justifies the conclusion because: [why]
2. [Second strongest] — this adds significant weight because: [why]
3. [Third strongest] — this reinforces the pattern because: [why]
...
N. [Weakest — include only if it adds something the others don't]
CUMULATIVE CASE: [Is the evidence collectively stronger than any single piece?]
Step 4: Address the Strongest Counterargument
Find the best argument against your conclusion and respond to it directly.
STRONGEST COUNTERARGUMENT:
[State it as strongly and fairly as possible — steelman it]
RESPONSE:
[Why the conclusion still holds despite this counterargument.
Options: the counterargument is factually wrong, it's true but
outweighed, it applies in theory but not in this context,
or it points to a real risk that can be mitigated.]
REMAINING COUNTERARGUMENTS:
- [Counter 2] — brief response: [why it doesn't change the conclusion]
- [Counter 3] — brief response: [why it doesn't change the conclusion]
HONEST CONCESSION: [What does the opposition get right?
Acknowledging this makes the rationale more credible.]
Step 5: Construct the Rationale
Combine everything into a persuasive but honest narrative.
RATIONALE:
[Opening: State the conclusion clearly.]
[Body: Present evidence in order of strength. Connect each piece
to the conclusion with explicit reasoning — don't just list facts,
show why they matter.]
[Counterargument: Acknowledge the strongest objection and explain
why it doesn't change the conclusion.]
[Close: Restate the conclusion with the accumulated weight of evidence
behind it. If appropriate, state what would change your mind.]
RATIONALE QUALITY CHECK:
- Persuasive: [Does it move the target audience?]
- Honest: [Does it acknowledge real weaknesses?]
- Complete: [Does it leave obvious questions unanswered?]
- Actionable: [Does it make clear what should happen next?]
GENERATION FAILURE CHECK:
- Pre-baked thesis: Would someone who had never read commentary on this topic reach the same conclusion from THIS evidence? If the rationale just arrives at the popular position, it's not a rationale — it's a dressed-up assumption. Rebuild from the specific evidence provided.
- Performed humility: Is there a "to be sure, there are limitations" paragraph? Delete it. If the rationale is the same or better, it was performance, not honesty. Real concessions change the shape of the argument — they cost something.
- Cached takes: Is any supporting argument a general principle rather than a claim derived from THIS specific evidence? ("Innovation drives growth" / "Data-driven decisions are better.") Replace with the actual reasoning chain from evidence to conclusion.
Integration
Use with:
/benf -> Generate benefit evidence to feed into the rationale/efrt -> Include effort estimates as part of the case/cand -> Build rationale for a candidate recommendation/aex -> Check assumptions embedded in the rationale
