critical-case-design

$npx mdskill add yogsoth-ai/de-anthropocentric-research-engine/critical-case-design

Maximizes inferential power by testing claims with critical cases

  • Evaluates claims by designing most- and least-likely cases
  • Uses parameter-space-mapping and extreme-value-generation tools
  • Applies Flyvbjerg methodology to identify favorable and unfavorable conditions
  • Reports confirmation or disconfirmation based on breakpoint and contradiction detection
SKILL.md
.github/skills/critical-case-designView on GitHub ↗
---
name: critical-case-design
description: "Flyvbjerg critical case methodology: select most-likely and least-likely cases to maximize inferential power."
type: strategy
used-by: [adversarial-stress-testing]
---

# Critical Case Design

## Tactics

- boundary-probing
- counterexample-heuristics

## Method

1. Identify the claim's scope and conditions
2. Design most-likely case: conditions maximally favorable to claim
3. Design least-likely case: conditions maximally unfavorable
4. If claim fails in most-likely case: strong disconfirmation
5. If claim holds in least-likely case: strong confirmation
6. Report inferential power of each test

## Budget

| Size | Critical cases designed | Evaluation depth |
|---|---|---|
| S | 2 (one most-likely, one least-likely) | Surface |
| M | 4 (two each) | Moderate |
| L | 6 (three each) | Deep |

## Orchestration

1. Dispatch `parameter-space-mapping` to identify conditions
2. Dispatch `extreme-value-generation` for favorable/unfavorable extremes
3. Dispatch `breakpoint-detection` to test claim at critical cases
4. Dispatch `contradiction-detection` to evaluate outcomes

## Subagents

- parameter-space-mapping
- extreme-value-generation
- breakpoint-detection
- contradiction-detection
More from yogsoth-ai/de-anthropocentric-research-engine