chain-roleplay-debate-synthesis

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Chain Roleplay → Debate → Synthesis

Chain Roleplay → Debate → Synthesis（角色扮演→辩论→合成法）

Workflow

工作流程

Copy this checklist and track your progress:

Roleplay → Debate → Synthesis Progress:
- [ ] Step 1: Frame the decision and identify roles
- [ ] Step 2: Roleplay each perspective authentically
- [ ] Step 3: Structured debate between viewpoints
- [ ] Step 4: Synthesize into coherent recommendation
- [ ] Step 5: Validate synthesis quality

Step 1: Frame the decision and identify roles

State the decision clearly as a question, identify 2-5 stakeholder perspectives or roles that have legitimate but competing interests, and clarify what a successful synthesis looks like. See Decision Framing for guidance on choosing productive roles.

Step 2: Roleplay each perspective authentically

For each role, articulate their position, priorities, concerns, and evidence. Genuinely advocate for each viewpoint without strawmanning. See Roleplay Guidelines for authentic advocacy techniques and use resources/template.md for complete structure.

Step 3: Structured debate between viewpoints

Facilitate direct clash between perspectives on key points of disagreement. Surface tensions, challenge assumptions, test edge cases, and identify cruxes (what evidence would change each perspective's mind). See Debate Structure for debate formats and facilitation techniques.

Step 4: Synthesize into coherent recommendation

Integrate insights from all perspectives into a unified decision that acknowledges tradeoffs, incorporates valid concerns from each viewpoint, and explains what's being prioritized and why. See Synthesis Patterns for integration approaches and resources/template.md for synthesis framework. For complex multi-stakeholder decisions, see resources/methodology.md.

Step 5: Validate synthesis quality

Check synthesis against resources/evaluators/rubric_chain_roleplay_debate_synthesis.json to ensure all perspectives were represented authentically, debate surfaced real tensions, synthesis is coherent and actionable, and no perspective was dismissed without engagement. See When NOT to Use This Skill to confirm this approach was appropriate.

复制此检查清单并跟踪进度：

Roleplay → Debate → Synthesis Progress:
- [ ] Step 1: Frame the decision and identify roles
- [ ] Step 2: Roleplay each perspective authentically
- [ ] Step 3: Structured debate between viewpoints
- [ ] Step 4: Synthesize into coherent recommendation
- [ ] Step 5: Validate synthesis quality

步骤1：明确决策并确定角色

将决策清晰地表述为一个问题，确定2-5个具有合理但相互竞争利益的利益相关者视角或角色，并明确成功的合成结果是什么样的。如需了解选择有效角色的指导，请参阅决策框架部分。

步骤2：真实演绎每个视角

针对每个角色，阐明他们的立场、优先级、担忧和依据。真实地为每个观点辩护，避免歪曲对方立场。如需了解真实演绎的技巧，请参阅角色扮演指南，完整结构可使用resources/template.md模板。

步骤3：不同观点间的结构化辩论

推动不同视角在关键分歧点上的直接碰撞。挖掘矛盾、挑战假设、测试边缘案例，并确定核心分歧点（即需要哪些证据才能改变各视角的立场）。如需了解辩论形式和引导技巧，请参阅辩论结构部分。

步骤4：整合为连贯的建议

将所有视角的见解整合为一个统一的决策，该决策需承认权衡关系、纳入各视角的合理担忧，并解释优先事项及其原因。如需了解整合方法，请参阅合成模式，合成框架可使用resources/template.md模板。针对复杂的多利益相关者决策，请参阅resources/methodology.md。

步骤5：验证合成质量

对照resources/evaluators/rubric_chain_roleplay_debate_synthesis.json检查合成结果，确保所有视角都得到真实呈现、辩论挖掘出了真实矛盾、合成结果连贯且可执行，且没有任何视角在未被充分讨论的情况下被忽视。请参阅何时不使用此方法确认该方法是否适用。

Decision Framing

决策框架

Choosing Productive Roles

选择有效的角色

Good role selection:

Competing interests: Roles have legitimate but different priorities (e.g., Speed Advocate vs. Quality Guardian)
Different expertise: Roles bring distinct knowledge domains (e.g., Engineer, Designer, Customer)
Value tensions: Roles represent incompatible values (e.g., Privacy Advocate vs. Personalization)
Stakeholder representation: Roles map to real decision-makers or affected parties

Typical role patterns:

Functional roles: Engineer, Designer, PM, Marketer, Finance, Legal, Customer
Archetype roles: Optimist, Pessimist, Risk Manager, Visionary, Pragmatist
Stakeholder roles: Customer, Employee, Investor, Community, Regulator
Value roles: Ethics Officer, Growth Hacker, Brand Guardian, Innovation Lead
Temporal roles: Short-term Thinker, Long-term Strategist

How many roles:

2 roles: Clean binary debate (build vs. buy, growth vs. profitability)
3 roles: Triadic tension (speed vs. quality vs. cost)
4-5 roles: Multi-stakeholder complexity (product strategy with eng, design, marketing, finance, customer)
Avoid >5: Becomes unwieldy, synthesis too complex

优质角色选择：

相互竞争的利益：角色拥有合理但不同的优先级（例如，速度倡导者vs质量守护者）
不同的专业知识：角色带来不同的知识领域（例如，工程师、设计师、客户）
价值观冲突：角色代表不相容的价值观（例如，隐私倡导者vs个性化推动者）
利益相关者代表性：角色对应真实的决策者或受影响方

典型角色模式：

职能角色：工程师、设计师、产品经理（PM）、营销人员、财务、法务、客户
原型角色：乐观主义者、悲观主义者、风险管理者、远见者、实用主义者
利益相关者角色：客户、员工、投资者、社区、监管机构
价值观角色：伦理专员、增长黑客、品牌守护者、创新负责人
时间维度角色：短期思考者、长期战略家

角色数量：

2个角色：清晰的二元辩论（自研vs外购、增长vs盈利）
3个角色：三方矛盾（速度vs质量vs成本）
4-5个角色：多利益相关者复杂性（涉及工程师、设计、营销、财务、客户的产品战略）
避免超过5个角色：会变得难以掌控，合成过程过于复杂

Framing the Question

明确决策问题

Strong framing:

"Should we prioritize X over Y?" (clear tradeoff)
"What's the right balance between A and B?" (explicit tension)
"Should we pursue strategy X?" (specific, actionable)

Weak framing:

"What should we do?" (too vague)
"How can we have our cake and eat it too?" (assumes false resolution)
"Who's right?" (assumes winner rather than synthesis)

优质的问题框架：

"我们应该优先考虑X还是Y？"（清晰的权衡）
"A和B之间的正确平衡是什么？"（明确的矛盾）
"我们应该推行策略X吗？"（具体、可执行）

劣质的问题框架：

"我们该怎么办？"（过于模糊）
"我们如何鱼与熊掌兼得？"（假设存在虚假的解决方案）
"谁是对的？"（假设存在赢家而非合成结果）

Roleplay Guidelines

角色扮演指南

Authentic Advocacy

真实辩护

Each role should:

State position clearly: What do they believe should be done?
Articulate priorities: What values or goals drive this position?
Surface concerns: What risks or downsides do they see in other approaches?
Provide evidence: What data, experience, or reasoning supports this view?
Show vulnerability: What uncertainties or limitations does this role acknowledge?

Avoiding strawmen:

❌ "The engineer just wants to use shiny new tech" (caricature)
✅ "The engineer values maintainability and believes new framework reduces technical debt"
❌ "Sales only cares about closing deals" (dismissive)
✅ "Sales is accountable for revenue and sees this feature as critical for competitive positioning"

Empathy without capitulation: You can deeply understand a perspective without agreeing with it. Each role should be the "hero of their own story."

每个角色应做到：

清晰陈述立场：他们认为应该采取什么行动？
阐明优先级：哪些价值观或目标驱动着这个立场？
提出担忧：他们认为其他方法存在哪些风险或弊端？
提供依据：哪些数据、经验或推理支持这个观点？
展现局限性：该角色承认哪些不确定性或限制？

避免歪曲立场：

❌ "工程师只是想用新奇的技术"（刻板印象）
✅ "工程师重视可维护性，认为新框架能减少技术债务"
❌ "销售只关心成交"（轻视对方）
✅ "销售对营收负责，认为这个功能对竞争定位至关重要"

共情但不妥协： 你可以深刻理解一个视角，但无需认同它。每个角色都应该是“自己故事里的主角”。

Perspective-Taking Checklist

视角代入检查清单

Debate Structure

辩论结构

Facilitating Productive Clash

推动有效的观点碰撞

Debate formats:

1. Point-Counterpoint

Role A makes case for their position
Role B responds with objections and counterarguments
Role A addresses objections
Repeat with Role B's case

2. Devil's Advocate

One role presents the "default" or "obvious" choice
Other roles systematically challenge assumptions and surface risks
Goal: Pressure-test before committing

3. Constructive Confrontation

Identify 3-5 key decision dimensions (cost, speed, risk, quality, etc.)
Each role articulates position on each dimension
Surface where perspectives conflict most

4. Crux-Finding

Ask each role: "What would need to be true for you to change your mind?"
Identify testable assumptions or evidence that would shift debate
Focus discussion on cruxes rather than rehashing positions

辩论形式：

1. 点对点辩论

角色A为自己的立场辩护
角色B提出反对意见和反驳
角色A回应反对意见
角色B再陈述自己的立场，重复上述流程

2. 魔鬼代言人

一个角色提出“默认”或“显而易见”的选择
其他角色系统性地质疑假设并挖掘风险
目标：在投入前测试方案的可行性

3. 建设性对抗

确定3-5个关键决策维度（成本、速度、风险、质量等）
每个角色阐明自己在每个维度上的立场
挖掘各视角冲突最激烈的地方

4. 核心分歧点挖掘

询问每个角色：“需要满足什么条件你才会改变立场？”
确定可测试的假设或能改变辩论走向的依据
将讨论重点放在核心分歧点上，而非重复立场

Questions to Surface Tensions

挖掘矛盾的问题

"What's the strongest argument against your position?"
"What does [other role] see that you might be missing?"
"Where is the irreducible tradeoff between your perspectives?"
"If you had to steelman the opposing view, what would you say?"
"What happens in edge cases for your approach?"
"What are you optimizing for that others aren't?"

"反对你立场的最有力论据是什么？"
"[其他角色]看到了哪些你可能忽略的点？"
"你们的视角之间存在哪些不可调和的权衡？"
"如果你要强化对方的观点，你会怎么说？"
"你的方法在边缘案例中会出现什么问题？"
"你在优化的目标是哪些其他人没有关注的？"

Red Flags in Debate

辩论中的警示信号

Premature consensus: Roles agree too quickly without surfacing real tensions
Talking past each other: Roles argue different points rather than engaging
Appeal to authority: "Because the CEO said so" rather than reasoning
False dichotomies: "Either we do X or we fail" without exploring middle ground
Unsupported claims: "Everyone knows Y" without evidence or reasoning

过早达成共识：角色们太快达成一致，没有挖掘出真实的矛盾
各说各话：角色们争论不同的点，而非真正交锋
诉诸权威：“因为CEO这么说”而非基于推理
虚假二分法：“要么做X要么失败”而不探索中间地带
无依据的主张：“大家都知道Y”而没有提供依据或推理

Synthesis Patterns

合成模式

Integration Approaches

整合方法

1. Weighted Synthesis

"We'll prioritize X, while incorporating safeguards for Y's concerns"
Example: "Ship fast (PM's priority), but with feature flags and monitoring (Engineer's concern)"

2. Sequencing

"First we do X, then we address Y"
Example: "Launch MVP to test market (Growth), then invest in quality (Engineering) once product-market fit is proven"

3. Conditional Strategy

"If condition A, do X; if condition B, do Y"
Example: "If adoption > 10K users in Q1, invest in scale; otherwise, pivot based on feedback"

4. Hybrid Approach

"Combine elements of multiple perspectives"
Example: "Build core in-house (control) but buy peripheral components (speed)"

5. Reframing

"Debate reveals the real question is Z, not X vs Y"
Example: "Debate about pricing reveals we need to segment customers first"

6. Elevating Constraint

"Identify the binding constraint both perspectives agree on"
Example: "Both speed and quality advocates agree engineering capacity is the bottleneck; synthesis is to hire first"

1. 加权合成

“我们将优先考虑X，同时纳入针对Y担忧的保障措施”
示例：“快速发布（产品经理的优先级），但需搭配功能开关和监控机制（工程师的担忧）”

2. 分阶段执行

“先做X，再处理Y”
示例：“推出MVP测试市场（增长目标），一旦验证产品市场契合度，再投入资源提升质量（工程目标）”

3. 条件性策略

“如果满足条件A，就做X；如果满足条件B，就做Y”
示例：“如果第一季度用户 adoption 超过10K，就投入资源做规模化；否则，根据反馈调整方向”

4. 混合方法

“结合多个视角的元素”
示例：“核心功能自研（掌控力），但外围组件外购（速度）”

5. 重新框架

“辩论揭示真正的问题是Z，而非X vs Y”
示例：“关于定价的辩论揭示我们首先需要对客户进行细分”

6. 聚焦约束条件

“确定双方视角都认同的关键约束”
示例：“速度和质量倡导者都认同工程产能是瓶颈；合成结果是先招聘人员”

Synthesis Quality Markers

合成质量标志

Strong synthesis:

✅ Acknowledges validity of multiple perspectives
✅ Explains what's being prioritized and why
✅ Addresses major concerns from each viewpoint
✅ Clear on tradeoffs being accepted
✅ Actionable recommendation
✅ Monitoring plan for key assumptions

Weak synthesis:

❌ "Let's do everything" (no prioritization)
❌ "X wins, Y loses" (dismisses valid concerns)
❌ "We need more information" (avoids decision)
❌ "It depends" without specifying conditions
❌ Vague platitudes without concrete next steps

优质合成结果：

✅ 承认多个视角的合理性
✅ 解释优先事项及其原因
✅ 回应每个视角的主要担忧
✅ 明确接受的权衡关系
✅ 可执行的建议
✅ 针对关键假设的监控计划

劣质合成结果：

❌ “我们要做所有事”（没有优先级）
❌ “X赢，Y输”（忽视合理担忧）
❌ “我们需要更多信息”（回避决策）
❌ “视情况而定”但不明确条件
❌ 模糊的陈词滥调，没有具体的下一步行动

Examples

示例

Example 1: Short-form Synthesis

示例1：简短合成

Decision: Should we rewrite our monolith as microservices?

Roles:

Scalability Engineer: We need microservices to scale independently and deploy faster
Pragmatic Engineer: Rewrite is 12-18 months with high risk; monolith works fine
Finance: What's the ROI? Rewrite costs $2M in eng time

Synthesis: Don't rewrite everything, but extract the 2-3 services with clear scaling needs (authentication, payment processing) as independent microservices. Keep core business logic in monolith for now. This addresses scalability concerns for bottleneck components (Scalability Engineer), limits risk and timeline (Pragmatic Engineer), and reduces cost to $400K vs. $2M (Finance). Revisit full migration if extracted services succeed and prove the pattern.

决策：我们是否应该将单体应用重写为微服务？

角色:

可扩展性工程师：我们需要微服务来实现独立扩展和更快部署
实用主义工程师：重写需要12-18个月，风险很高；当前单体应用运行良好
财务：投资回报率是多少？重写需要200万美元的工程人力成本

合成结果: 不重写整个应用，但将2-3个有明确扩展需求的服务（认证、支付处理）拆分为独立的微服务。核心业务逻辑暂时保留在单体应用中。这样既解决了瓶颈组件的可扩展性问题（可扩展性工程师的需求），又限制了风险和时间线（实用主义工程师的担忧），还将成本从200万美元降至40万美元（财务的需求）。如果拆分的服务取得成功并验证了模式，再重新考虑全面迁移。

Example 2: Full Analysis

示例2：完整分析

For a complete worked example with detailed roleplay, debate, and synthesis, see:

resources/examples/build-vs-buy-crm.md - Sales, Engineering, Finance debate CRM platform decision

如需查看包含详细角色扮演、辩论和合成结果的完整示例，请参阅：

resources/examples/build-vs-buy-crm.md - 销售、工程、财务团队关于CRM平台的决策辩论

When NOT to Use This Skill

何时不使用此方法

Skip roleplay-debate-synthesis when:

❌ Single clear expert: If one person has definitive expertise and others defer, just ask the expert ❌ No genuine tension: If stakeholders actually agree, debate is artificial ❌ Values cannot be negotiated: If ethical red line, don't roleplay the unethical side ❌ Time-critical decision: If decision must be made in minutes, skip full debate ❌ Implementation details: If decision is "how" not "whether" or "what", use technical collaboration not debate

Use simpler approaches when:

✅ Decision is straightforward with clear data → Use decision matrix or expected value
✅ Need creative options not evaluation → Use brainstorming not debate
✅ Need detailed analysis not perspective clash → Use analytical frameworks
✅ Implementation planning not decision-making → Use project planning not roleplay

请勿使用角色扮演→辩论→合成法的场景：

❌ 单一明确专家：如果一个人拥有绝对专业知识，其他人都听从其意见，直接咨询该专家即可 ❌ 无真实矛盾：如果利益相关者实际意见一致，辩论是人为的 ❌ 价值观不可协商：如果涉及伦理红线，不要演绎不道德的一方 ❌ 时间紧迫的决策：如果必须在几分钟内做出决策，跳过完整辩论 ❌ 实施细节：如果决策是关于“如何做”而非“是否做”或“做什么”，使用技术协作而非辩论

以下场景使用更简单的方法：

✅ 决策简单且有明确数据支撑 → 使用决策矩阵或期望值法
✅ 需要创意方案而非评估 → 使用头脑风暴而非辩论
✅ 需要详细分析而非视角碰撞 → 使用分析框架
✅ 实施规划而非决策制定 → 使用项目规划而非角色扮演

Advanced Techniques

高级技巧

For complex multi-stakeholder decisions, see resources/methodology.md for:

Multi-round debates (iterative refinement of positions)
Audience-perspective shifts (how synthesis changes for different stakeholders)
Facilitation anti-patterns (how debates go wrong)
Synthesis under uncertainty (when evidence is incomplete)
Stakeholder mapping (identifying who needs to be represented)

针对复杂的多利益相关者决策，请参阅resources/methodology.md了解：

多轮辩论（迭代优化立场）
受众视角转换（针对不同利益相关者的合成结果调整）
引导反模式（辩论容易出现的问题）
不确定性下的合成（当依据不完整时）
利益相关者映射（确定需要代表的对象）

Resources

资源

resources/template.md - Structured template for roleplay → debate → synthesis analysis
resources/methodology.md - Advanced facilitation techniques and debate formats
resources/examples/ - Complete worked examples across domains
resources/evaluators/rubric_chain_roleplay_debate_synthesis.json - Quality assessment rubric (10 criteria)

resources/template.md - 角色扮演→辩论→合成分析的结构化模板
resources/methodology.md - 高级引导技巧和辩论形式
resources/examples/ - 跨领域的完整示例
resources/evaluators/rubric_chain_roleplay_debate_synthesis.json - 质量评估 rubric（10项标准）