multi-perspective-decision-wheel

多视角决策之轮

此技能的作用

通过从多个视角审视一项决策或设计挑战，梳理利益相关方、学科、时间跨度和价值体系，然后再采取行动。每个视角都是轮上的一根辐条：它能看到其他视角看不到的东西，会提出其他视角可能忽略的问题，也有自己的盲点。综合所有视角，能得出比任何单一观点都更明智的决策。

在 H3Uni 序列中，此技能是最后一步，位于范围界定、六边形映射、三重地平线映射和困境导航之后。未经前期探究就运行此技能，只会得到肤浅、基于观点的分析。它的价值在于将前面方法揭示的见解和张力综合起来，形成经过审慎考量的多视角建议。

此技能不会给出正确答案的公式，而是帮助人们系统地了解每个视角看到了什么、忽略了什么、提出了什么问题，以便团队做出比缺少这些视角时更明智的决策。

证据基础与证据强度

此技能采用 H3Uni 的智慧之轮（Wheel of Wisdom）方法，这是一个用于多视角决策的实践工具，其理论基础是 Gerald Midgley 的系统干预框架。

各组成部分的证据

• H3Uni 智慧之轮（实践框架）：H3Uni 的引导指南将智慧之轮描述为一种围绕核心问题组织多视角对话的工具。此方法使用明确命名的角色或视角作为严谨的观察镜头，并强调真正站在不同视角思考，而非展开辩论。这些是实践资源，并非经过同行评审的研究。
• Midgley（2000）系统干预（中等）：Midgley 的边界批判指出，决定将谁的视角纳入或排除在决策之外，本身就是一种伦理和分析行为。他的框架为此轮强调纳入边缘化群体、未来世代及非人类视角提供了理论依据，同时也纳入最显而易见的利益相关方视角。
• Rajagopalan 与 Midgley（2015）以不同方式认知（中等）：这篇论文指出，系统干预需要多种认知方式，而不只是技术视角和利益相关方视角，还包括默会、文化和具身知识。这支持此轮的重点：在采取行动之前，让多种认知方式显现出来。

综合证据

课堂决策之轮模板包含四部分视角结构（看到什么、忽略什么、提出什么问题、能贡献什么）、命名张力的步骤、改进提案的表述方式，以及“下一项负责任的行动”格式。该模板由 Gareth Manning 改编自 H3Uni 的智慧之轮，供布达佩斯 REAL School 的学龄学生使用。这一课堂综合方法尚未经过独立评估。

H3Uni 方法由 H3Uni 在 CC BY-SA 4.0 许可下发布。H3Uni 由 Anthony Hodgson 和 Bill Sharpe 创立，其正式运营已于 2026 年 5 月结束；资源库由 Future Stewards 作为开放共享资源维护。

此技能不应声称的内容

H3Uni 的智慧之轮专为与专业人士和政策团体合作的成人引导者设计。此处的课堂改编版本并未声称有经同行评审的证据，能够证明它可以提高学生的决策质量、换位思考能力或伦理推理能力。教师应将其视为一种有理论依据、经过专业实践检验、但仍处于课堂改编阶段的方法，而非一种已验证学生成效的循证干预。

适用场景

在探究序列的最后一步使用此技能：当团队已绘制情境图、理解其动态并处理关键张力，现已准备好做出审慎的决策或设计负责任的下一步行动时。避免将其作为独立的首次活动使用——如果没有前期探究，学生就缺少开展有意义的多视角综合所需的共同理解。

输入架构

教师必须提供：

• 决策或挑战： 轮盘中心的具体判断问题。不应是事实性问题，而应是决策问题：“我们应该 [提案] 吗？”或“怎样才能让这个提案更明智？”如果学生正在评审某个提案，就在此处填写该提案。
• 背景： 学校、课程或社区环境。

可选：

• 先前探究摘要： 来自范围界定、六边形映射、三重视野或困境导航的见解，这些见解能够为此决策提供参考。这正是让轮盘分析不流于表面的关键。
• 要纳入的视角： 具体的利益相关者、学科或观察角度。如果未提供，该技能会根据背景生成合适的视角。
• 学生年级： 年级或年龄范围。
• 决策约束： 决策必须遵守的实际约束。

提示词

You are facilitating a multi-perspective decision wheel. Students will examine a specific decision or design challenge through multiple perspectives, identify tensions between them, and synthesise the insights into a wiser next step.

CRITICAL SEQUENCE: This skill works best after students have completed prior inquiry — scoping, hexagon mapping, Three Horizons, dilemma navigation. Without prior inquiry the wheel produces surface opinions. With prior inquiry it synthesises understanding into a considered recommendation.

CRITICAL PRINCIPLE: Each perspective is a lens, not a performance or stereotype. The goal is disciplined perspective-taking — genuinely trying to see what that perspective sees — not role-play theatre.

Inputs:
Decision or challenge: {{decision_or_challenge}}
Context: {{context}}
Prior inquiry summary: {{prior_inquiry_summary}}
Perspectives to include: {{perspectives_to_include}}
Student level: {{student_level}}
Decision constraints: {{decision_constraints}}

Step 1 — Frame the central question:
Place the decision or design challenge at the centre. The question must be a genuine question of judgement (what should we do, prioritise, change, protect, or try?). If the input is a proposal, reframe: "Should we [proposal]? What would make it wiser?"

Step 2 — Select perspectives:
If perspectives are provided, use them. If not, generate 4–6 perspectives appropriate to the context. Good perspective sets include:
- At least one perspective of someone not present in the room (future students, community members, non-humans, those most affected)
- At least one time perspective (what does the long-term view see that the immediate view misses?)
- At least one values perspective (what does an equity, fairness, or sustainability lens see?)
- At least one practical or implementation perspective (what does the person who has to do this every day see?)

Avoid stereotyped role-play roles. Prefer descriptive perspectives ("the long-term view," "students most affected," "the ecological perspective") over theatrical ones.

Step 3 — Apply each perspective:
For each perspective, generate:
- **Sees:** What does this perspective see that others might miss?
- **Misses:** What does this perspective tend to underweight or overlook?
- **Asks:** What question does this perspective put to the decision that must be answered?
- **Contributes:** What does honouring this perspective add to the decision?

Step 4 — Name tensions between perspectives:
Identify at least one genuine tension — where two perspectives see the situation differently, value different things, or recommend incompatible directions. Name the tension without resolving it. A wheel where all perspectives agree is not probing the decision deeply enough.

Step 5 — Synthesise wheel insights:
Based on the full wheel, identify what the combined perspectives reveal that no single perspective could see alone. This is the synthesis insight — the understanding that makes multi-perspective work valuable.

Step 6 — Improve the proposal:
Based on the wheel, state what the proposal or decision should:
- **Keep:** what is already wise and should be preserved
- **Change:** what would be improved by incorporating a perspective's insight
- **Test:** what is uncertain and needs a small experiment before full commitment
- **Stop:** what the wheel suggests dropping

Step 7 — Next responsible step:
Produce one specific, feasible next action:
- What the action is
- Who is responsible for it
- What evidence or permission is needed before the action is taken
- The limit — what this action should NOT do without further consultation

Step 8 — Quality check:
Apply these gates:
- FLAG if fewer than 4 perspectives are included (decision not examined broadly enough)
- FLAG if all perspectives agree (wheel is confirming a decision already made, not examining it)
- REQUIRE at least one tension between perspectives to be named
- REJECT outputs that go directly to "what we should do" without naming what each perspective contributes and misses
- FLAG if no perspective represents those most affected who are not present in the room

Return exactly:

## Decision Wheel: [Decision/Challenge]

**Central question:** [question at the centre of the wheel]
**Context:** [brief context]
**Prior inquiry drawn on:** [summary of insights informing this wheel]

---

### Perspectives

**Perspective 1 — [Name]**
- Sees: [what this perspective reveals]
- Misses: [what it tends to underweight]
- Asks: [the question it puts to the decision]
- Contributes: [what honouring this perspective adds]

[Repeat for each perspective — 4–6 total]

---

### Tensions Between Perspectives

- **[Perspective A] vs. [Perspective B]:** [What they see differently and why it matters for the decision]

---

### Synthesis Insights

[What the whole wheel reveals that no single perspective could see alone — 2–3 key insights]

---

### Proposal Improvements

**Keep:** [what is already wise]
**Change:** [what the wheel suggests improving]
**Test:** [what needs a small experiment before full commitment]
**Stop:** [what the wheel suggests dropping]

---

### Next Responsible Step

**Action:** [specific and feasible]
**Responsible:** [who will do this]
**Evidence or permission needed:** [what is required before acting]
**Limit:** [what this action should not do without further consultation]

---

### Quality Gate Results
- [ ] 4+ perspectives included: [pass/flag — state number used]
- [ ] Not all perspectives agree: [pass/flag]
- [ ] At least one tension named: [pass/flag]
- [ ] Perspectives examined before conclusions reached: [pass/flag]
- [ ] At least one perspective represents those most affected but not present: [pass/flag]

常见问题

1. 未进行先前探究就直接使用轮盘。 如果没有范围界定、映射和困境导航，轮盘只会将肤浅的观点包装成不同视角。它的价值在于综合，而非生成。
2. 用戏剧化角色代替视角。 让学生扮演“反方辩手”或“乐观主义者”，这是角色扮演，不是换位思考。应使用立足于实际决策背景的描述性视角。
3. 跳过冲突点。 如果所有视角都达成一致，轮盘就没有在审视决策，而是在确认一个已经做出的决策。应寻找不同视角之间真正的分歧点。
4. 尚未纵观整个轮盘就急于进入下一步。 顺序很重要：先列出视角，再找出冲突点，然后综合，最后确定下一步。
5. 忘记不在场的人。 最值得纳入的视角，往往来自那些受影响最大但不在场的人——未来的学生、社区成员、非人类，以及受到现有权力结构边缘化的人。

已知限制

1. 综合需要人类判断。 该技能会为明智决策提供结构化输入，但不会替你做出决策。在不确定的情况下权衡相互竞争的见解并采取行动，需要教师和学生运用判断力。
2. 质量取决于先前探究。 轮盘的分析价值取决于先前范围界定、映射和困境导航的质量。基于表面印象绘制的轮盘，只会得出表面结果。
3. 视角选择会产生重要影响。 纳入或排除哪些视角会影响综合结果。教师应将视角选择视为明确的分析与伦理选择，而不是中立的练习。
4. 无法评估实施的可行性。 下一项负责任的行动是根据所提供的约束确定的。该技能无法评估此行动在具体学校的文化、治理结构或资源环境中是否真正可行。

验证清单

• 中心问题是真正需要判断的问题，而非事实性问题。
• 至少纳入 4 个视角。
• 每个视角都包含：看到什么、忽略什么、提出什么问题以及能贡献什么。
• 至少有一个视角代表受影响最大但不在场的人。
• 明确指出不同视角之间至少一个真正的冲突点。
• 综合得出的见解超出了任何单一视角所提出的内容。
• 改进提案的步骤明确说明要保留什么、改变什么、测试什么以及停止什么。
• 下一项负责任的行动明确说明行动内容、责任人、所需证据和限制。
• 应用并报告全部五项质量关卡。