seesaw

name	seesaw
description	Safety gate. Weighs whether a conclusion (positioning gap, GTM recommendation, market-entry verdict) is actually balanced against evidence on both sides, not just the evidence that supports it. Run after `decide` or `execute` produces a verdict, before `critic`.

Seesaw — Balance Check on Conclusions

A conclusion can be fully sourced (passes critic) and still be one-sided — every citation supports "enter this market" and none were sought that might argue against it. This gate specifically checks for that imbalance.

Process

1. Identify the conclusion under review — e.g., "the positioning gap is defensible," "channel X is the right primary channel," "enter this market."
2. List the evidence FOR the conclusion as already documented.
3. Actively search for evidence AGAINST it that wasn't yet considered:
   • Why might the "gap" exist because competitors tried and failed there?
   • Why might the chosen channel underperform for this specific ICP despite benchmarks?
   • What would a competitor's own GTM analyst say is wrong with this plan?
4. Weigh both sides explicitly. Do not average them into mush — state which side is stronger and why, in one paragraph a skeptic would accept.
5. If the against-evidence is stronger or comparable, the conclusion must be revised or downgraded from "recommended" to "worth testing cautiously" — do not force a confident verdict onto mixed evidence.

Write to gtm/SEESAW-CHECK.md (or append if reviewing multiple conclusions):

---
created: {ISO-8601}
---

# Seesaw Check: {conclusion under review}

## For
- {evidence point} — [ref]
- ...

## Against
- {evidence point actively sought, not assumed absent} — [ref or reasoning]
- ...

## Weighing
{one paragraph: which side wins and why, or why the verdict must be
downgraded to "test cautiously"}

## Final Calibrated Verdict
{the original conclusion, or its downgraded/revised version}

Anti-Rationalization

• "I didn't find counter-evidence" is different from "I looked for counter-evidence and found none" — do the latter, and say which happened.
• A GTM plan that only cites data supporting the chosen channel, with no search for competitor failures in that channel, hasn't been seesaw-checked — it's been confirmed.
• Downgrading a verdict is not failure — presenting an overconfident verdict that later fails in market is the actual failure this gate prevents.

Evidence

Complete when at least one specific counter-argument was actively sought (not assumed absent) for each major conclusion, and the final verdict reflects the actual balance rather than the initial framing.

Red Flags

• "Against" section left empty with no search described.
• Weighing section that restates the "For" evidence without engaging the "Against" evidence at all.
• Final verdict identical to the pre-check conclusion despite strong counter-evidence being listed.