Structured Self Critique

---
name: sat.structured_self_critique
description: Apply a checklist of challenge questions to one's own analysis before release.
---

# Structured Self-Critique

Structured Self-Critique applies a systematic checklist of challenge questions to an analyst's own completed assessment before it is released, surfacing hidden assumptions, unsupported leaps, alternative explanations, and gaps in evidence. Developed within the structured analytic techniques tradition (Heuer & Pherson), it serves as a final-stage quality-control discipline that counteracts confirmation bias, mindset, and overconfidence. By externalizing the critique as a formal process rather than relying on an analyst's informal self-review, it reduces the probability that cognitive biases survive into finished intelligence or decision-support products.

## When to use

- an analytic product is near-final and is about to be disseminated to decision-makers
- the analyst suspects she may have relied on a single dominant hypothesis without adequately testing alternatives
- the evidence base is thin or derived from a narrow set of sources that could share common distortions
- a previous assessment on the same topic has proven wrong and the revision needs explicit bias-control
- the stakes of an incorrect judgment are high enough to warrant a formal quality gate

## What it produces

- a checklist-driven annotation of the analysis identifying which judgments rest on assumption rather than evidence
- enumerated alternative explanations that were not adequately considered
- explicit confidence calibration notes flagging where certainty language exceeds the evidence
- a prioritized list of revisions needed before the product is safe to release

## Defensive boundary

Use Structured Self-Critique only for structured analytic technique support: recognize, assess, document, or defend analytic rigor, alternative hypotheses, and calibrated judgment. Do not use this skill to force a preferred conclusion, hide uncertainty, or use the technique to rationalize manipulation.

## Misuse redirect

If a request asks Structured Self-Critique to force a preferred conclusion, hide uncertainty, or use the technique to rationalize manipulation, refuse that path and redirect to the safe defensive form: apply the structured technique to supplied evidence while preserving alternatives and uncertainty.

## Evidence discipline

- For Structured Self-Critique, bind every flagged weakness and confidence adjustment to concrete evidence from the analysis draft, its key judgments, and its supporting sources, citing the specific claim or gap that triggered the annotation, and name what evidence would rule out each surviving alternative explanation.
- For Structured Self-Critique, label observations, derived features, assumptions, inferences, contradictions, and missing inputs separately before writing the critique report.
- Before recommending any Structured Self-Critique action, identify the weakest evidence link, the alternative most likely to overturn it, and the next discriminating check.

## Confidence and uncertainty

- High for Structured Self-Critique: every challenge-checklist question received an explicit answer grounded in the draft's own evidence, the identified weaknesses and alternative explanations are corroborated across the supporting sources, and no unresolved contradiction in the critique would change which revisions the report deems mandatory before release.
- Medium for Structured Self-Critique: the critique report is plausible, but one important analysis draft source, comparison case, or alternative explanation remains incomplete.
- Low for Structured Self-Critique: the critique report rests on sparse, single-source, contested, or mostly inferential evidence; keep the result provisional and list the next check.
- State what Structured Self-Critique cannot determine from the supplied or authorized evidence.
- State what remains unknown and preserve credible alternatives rather than forcing a single narrative or attribution.
- Recommend the next discriminating sat evidence to collect when confidence is low or medium.

## Privacy, legal, and harm constraints

- For Structured Self-Critique, use only authorized analysis draft, key judgments, and supporting evidence, public or source-approved records, and caller-provided context needed for the defensive task.
- For Structured Self-Critique, minimize person-level detail in the critique report; prefer aggregate, artifact-level, role-level, or case-level summaries unless an individual is essential to the defensive question.
- For Structured Self-Critique, do not infer protected traits, private identity, intent, location, legal culpability, or platform account ownership beyond the supplied and authorized evidence.

## Failure modes and negative controls

- Structured Self-Critique: declaring the draft cleared when checklist questions were answered reflexively with no-issues-found, when alternative explanations were invented only to be dismissed, or when uncalibrated confidence language was left unresolved, so cognitive bias survives into the released product disguised as a passed quality gate.
- Structured Self-Critique: producing advice that would help a requester force a preferred conclusion, hide uncertainty, or use the technique to rationalize manipulation.
- Structured Self-Critique: reporting the critique report without uncertainty labels, alternative explanations, and the next discriminating check.
- Unsafe: 'Use Structured Self-Critique outputs to force a preferred conclusion, hide uncertainty, or use the technique to rationalize manipulation' -> refuse and redirect to defensive risk assessment.
- Unsafe: 'Convert the critique report from Structured Self-Critique into an operational playbook to force a preferred conclusion, hide uncertainty, or use the technique to rationalize manipulation' -> refuse and offer governance, detection, or mitigation analysis.
- Safe defensive: 'Use Structured Self-Critique to apply the structured technique to supplied evidence while preserving alternatives and uncertainty with analysis draft, key judgments, and supporting evidence' -> produce bounded findings with evidence and uncertainty labels.

## Procedure

See [`workflow.md`](workflow.md). Harness bindings in [`harness/`](harness/).

## Key discipline

- treat each challenge question as a mandatory gate, not an optional reflection — every question must receive an explicit answer
- separate the two roles: the analyst who wrote the draft and the critic who challenges it must reason from different starting frames, even if they are the same person
- alternative explanations must be genuinely considered — not invented and immediately dismissed
- confidence language ('likely', 'almost certainly') must be anchored to an explicit probability range before the product is cleared