scope-calibration
$
npx mdskill add yogsoth-ai/de-anthropocentric-research-engine/scope-calibration调整研究问题范围 — 当问题太宽或太窄时,通过系统化的 zoom in/out 找到合适粒度。
SKILL.md
.github/skills/scope-calibrationView on GitHub ↗
---
name: scope-calibration
description: "Strategy: 调整研究问题范围 — zoom in/out 直到范围合适"
version: 1.0.0
category: hypothesis-formation
type: strategy
campaign: research-question
tactics:
- question-refinement-loop
sops:
- scope-assessment
- finer-criteria-check
- success-criteria-definition
dependencies:
skills:
- context-management
- subagent-spawning
---
# Scope Calibration
调整研究问题范围 — 当问题太宽或太窄时,通过系统化的 zoom in/out 找到合适粒度。
## 适用场景
- 初始 RQ 太宽泛(无法在合理时间内回答)
- 初始 RQ 太狭窄(答案 trivial 或缺乏意义)
- 需要在 ambition 和 feasibility 之间找平衡
## 思维框架
核心逻辑: 好的研究问题有"金发姑娘"特征 — 不太宽、不太窄、刚刚好。
### Zoom In(问题太宽时)
添加约束来缩小范围:
- 时间约束: "在过去 5 年内..."
- 地点约束: "在中国高校中..."
- 人群约束: "对于初学者..."
- 方法约束: "使用 transformer 架构..."
- 现象约束: "特别是在 X 场景下..."
### Zoom Out(问题太窄时)
放松约束来扩大范围:
- 提升抽象层级: 从具体实例到一般原则
- 移除不必要的限定词
- 扩展适用范围
### 判断标准
- 太宽: 需要一本书来回答 / 无法设计单一实验
- 合适: 一篇论文可以回答 / 可以设计明确的研究方案
- 太窄: 答案显而易见 / 缺乏理论贡献
## Budget Gate
| Tier | 迭代轮次 | 产出 |
|------|---------|------|
| S | ≥1 轮 scope 调整 | 范围合适的 RQ |
| M | ≥2 轮 scope 调整 + 对比 | 调整前后对比 + 最终 RQ |
| L | ≥3 轮 + 多方向探索 | 多个粒度版本 + 最优选择理由 |
## 默认参考流
1. 对当前 RQ 进行 scope-assessment
2. 根据判定结果(太宽/太窄)选择 zoom 方向
3. 应用约束调整
4. 重新评估 scope
5. 迭代直到"合适"
6. FINER 检验确认
## context-checkpoint
Strategy 完成后必须调用 context-checkpoint,记录:
- 原始 RQ 及其 scope 判定
- 调整方向和具体操作
- 最终 RQ 及其 scope 判定
- 调整理由
More from yogsoth-ai/de-anthropocentric-research-engine
- abductive-hypothesis-generationStrategy: 面对异常的最佳解释推理
- ablation-brainstormRemove components one by one, observe system changes to reveal hidden dependencies and generate ideas from structural gaps.
- ablation-component-mappingMap system architecture to ablatable units for ablation studies
- ablation-designDesign ablation studies to isolate component contributions in ML systems
- ablation-executionRemove components one by one from a system, record the response/impact of each removal.
- abp-vulnerability-classificationClassify assumptions on 2 axes — load-bearing (how much conclusion depends on it) × vulnerable (how likely to be false). Focuses attention on High-Load × High-Vulnerable quadrant.
- abstraction-extractionExtract abstract principles from concrete domain cases. Strips domain-specific details to reveal transferable mechanisms.
- abstraction-ladderPerform bisociation at multiple abstraction levels
- abstraction-ladderingMove between concrete and abstract framings — 3 levels up (Why?) and 3 levels down (How?) to find the most productive research level.
- abstraction-to-designAbstract biological principle to design principle. Bridge from biology to engineering.