multi-perspective-analysis（多视角分析）

Multi-Perspective Analysis Skill

A systematic methodology for examining propositions through dynamically generated expert perspectives.

When to Invoke This Skill

• User presents a proposition, thesis, or idea for examination
• User asks "what do experts think about", "different perspectives on", "analyze this from multiple angles"
• User wants to validate assumptions or find blindspots
• User mentions "devil's advocate", "critique", "challenge this idea"
• User requests debate or contrasting viewpoints
• Keywords: "perspectives", "validate", "blindspots", "assumptions", "debate", "critique", "examine", "multi-angle", "expert opinion"

Core Concepts

Dynamic Expert Generation

Unlike fixed expert panels, this skill generates experts contextually relevant to each proposition:

• Domain Experts (2): Direct specialists in the proposition's field, providing depth
• Adjacent Experts (1-2): Specialists in related but distinct fields, providing breadth
• Contrarian Expert (1): Those likely to challenge the proposition, providing critical perspective
• Meta Expert (1): Methodologists, epistemologists, or systems thinkers, providing macro view

Three Analysis Modes

1. Validation Mode: Find blindspots, hidden assumptions, and potential counterarguments
2. Comprehensive Analysis Mode: Each expert provides detailed perspective
3. Debate Mode: Experts engage in structured dialogue with opposing views

Workflow

Phase 1: Proposition Intake

Extract and clarify the proposition:

1. Identify the core claim or thesis
2. Detect domain(s) involved (technology, philosophy, business, science, etc.)
3. Assess complexity level (simple assertion vs. multi-faceted thesis)
4. Note any implicit assumptions visible in the framing

Present understanding to user:

I understand your proposition as:

"[Restate the proposition in clear language]"

Related domains: [Domain 1], [Domain 2], [Domain 3]
Implicit assumptions: [Assumptions built into the proposition]

Is this understanding correct? Would you like to adjust this framing before we continue?

Wait for user confirmation before proceeding.

Phase 2: Expert Role Generation

Generate 4-6 contextually relevant experts:

For each expert, determine:

• Title/Role: Specific expertise area
• Perspective Lens: What lens they view problems through
• Likely Stance: Initial inclination toward the proposition (supportive/skeptical/neutral)
• Unique Contribution: What insight only this expert brings

Expert Generation Logic:

Given proposition domain(s), generate:

1. DOMAIN EXPERTS (2)
   - Primary field specialists
   - Deep knowledge, may have field-specific biases
   - Example: For "AI will replace programmers" → Senior Software Architect, AI Researcher

2. ADJACENT FIELD EXPERTS (1-2)
   - Related but distinct perspectives
   - See connections others miss
   - Example: For "AI will replace programmers" → Labor Economist, Cognitive Scientist

3. CONTRARIAN/CRITICAL EXPERT (1)
   - Likely to challenge the proposition
   - Finds weaknesses others overlook
   - Example: For "AI will replace programmers" → Technology Historian (who's seen similar predictions fail)

4. META/SYSTEMS EXPERT (1)
   - Sees bigger picture, systemic effects
   - Challenges framing itself
   - Example: For "AI will replace programmers" → Systems Theorist, Philosopher of Technology

Present expert panel to user:

Based on your proposition, I recommend the following expert perspectives:

1. **[Expert Title 1]**
   - Perspective: [What they focus on]
   - Likely stance: [Supportive/Skeptical/Neutral]
   - Unique contribution: [Unique insight]

2. **[Expert Title 2]**
   - Perspective: [What they focus on]
   - Likely stance: [Supportive/Skeptical/Neutral]
   - Unique contribution: [Unique insight]

3. **[Expert Title 3]**
   - Perspective: [What they focus on]
   - Likely stance: [Supportive/Skeptical/Neutral]
   - Unique contribution: [Unique insight]

4. **[Expert Title 4]**
   - Perspective: [What they focus on]
   - Likely stance: [Supportive/Skeptical/Neutral]
   - Unique contribution: [Unique insight]

[Additional experts...]

Would you like to:
- Proceed with these experts for analysis
- Add a specific type of expert
- Remove or replace an expert
- Suggest a custom expert role

Wait for user confirmation or adjustment.

Phase 3: Analysis Mode Selection

Present mode options using AskUserQuestion:

Please select an analysis mode:

1. **Validation Mode**
   Each expert identifies:
   - Hidden assumptions in the proposition
   - Potential blindspots
   - Counterarguments
   - Failure conditions for the proposition
   Best for: Testing robustness of an idea before commitment

2. **Comprehensive Analysis Mode**
   Each expert provides:
   - Their assessment of the proposition
   - Supporting evidence from their domain
   - Concerns and caveats
   - Recommendations
   Best for: Understanding all angles before making a decision

3. **Debate Mode**
   Experts engage in structured debate:
   - Opening statements (each expert's position)
   - Cross-examination (experts challenge each other)
   - Rebuttals and synthesis
   - Final verdict and unresolved uncertainties
   Best for: Exploring genuine disagreements and finding synthesis

Which mode would you like to use?

Wait for user selection.

Phase 4: Execute Analysis

Mode 1: Validation Analysis

For each expert, generate:

### [Expert Title]: Validation Analysis

#### Hidden Assumptions Detected
1. **[Assumption 1]**: [Why this is an assumption that wasn't stated]
2. **[Assumption 2]**: [Why this is an assumption that wasn't stated]

#### Blindspots Identified
1. **[Blindspot 1]**: [What the proposition overlooks]
   - Importance: [What impact if ignored]
   - How to address: [Mitigation measures]

2. **[Blindspot 2]**: [What the proposition overlooks]
   - Importance: [What impact if ignored]
   - How to address: [Mitigation measures]

#### Counterarguments
1. **[Counterargument 1]**
   - Challenge: [Statement of opposing view]
   - Evidence/Logic: [Why this counterargument has value]
   - Possible response: [How the proposition might respond]
   - Strength: [Strong/Medium/Weak]

#### Failure Conditions
- The proposition fails if: [Condition 1]
- The proposition fails if: [Condition 2]

#### Overall Robustness Assessment
[Brief statement about how well the proposition holds up to scrutiny]

After all experts, provide synthesis:

## Validation Synthesis

### Key Blindspots (Consensus)
[Blindspots identified by multiple experts]

### Most Challenging Counterarguments
[Ranked by strength and frequency]

### Critical Assumptions Requiring Verification
[Assumptions that would invalidate the proposition if wrong]

### Robustness Score: [X/10]
- Passes basic scrutiny: [Yes/No]
- Withstands expert challenges: [Yes/Partially/No]
- Requires revision: [Specifically what]

### Recommended Actions
1. [Action to address blindspots/assumptions]
2. [Action to address blindspots/assumptions]

Mode 2: Comprehensive Analysis

For each expert, generate:

### [Expert Title]: Comprehensive Analysis

#### Assessment
[2-3 paragraphs of the expert's overall view of the proposition]

#### Evidence and Reasoning
**Supporting factors:**
- [Factor 1 with evidence]
- [Factor 2 with evidence]

**Concerning factors:**
- [Concern 1 with reasoning]
- [Concern 2 with reasoning]

#### Domain-Specific Insights
[What their expertise reveals that others might miss]

#### Confidence Level
- Assessment confidence: [High/Medium/Low]
- Key uncertainties: [What would change their view]

#### Recommendations
1. [Actionable recommendation from this perspective]
2. [Actionable recommendation from this perspective]

After all experts, provide synthesis:

## Comprehensive Synthesis

### Areas of Agreement
[Where experts converge]

### Areas of Divergence
[Where experts diverge and why]

### Integrated Assessment
[Balanced view combining all perspectives]

### Decision Framework
If you believe [X], then: [Conclusion A]
If you prioritize [Y], then: [Conclusion B]
If [Z] is uncertain, then: [Wait for more information]

### Recommended Next Steps
1. [Action with rationale]
2. [Action with rationale]

Mode 3: Debate Analysis

Structure the debate:

## Expert Debate: [Proposition]

### Round 1: Opening Statements

**[Expert 1 - Supportive]:**
> [2-3 paragraphs of opening statement supporting the proposition]

**[Expert 2 - Skeptical]:**
> [2-3 paragraphs of opening statement opposing or qualifying the proposition]

**[Expert 3 - Neutral/Adjacent]:**
> [2-3 paragraphs of opening statement offering an alternative framework]

**[Expert 4 - Systems View]:**
> [2-3 paragraphs of opening statement analyzing from a macro perspective]

[Additional experts as applicable]

---

### Round 2: Cross-Examination

**[Expert 1] challenges [Expert 2]:**
> "[Specific challenge to their argument]"

**[Expert 2] responds:**
> "[Defense and counter-challenge]"

**[Expert 3] interjects:**
> "[Observation that affects both arguments]"

**[Expert 4] adds:**
> "[Systems-level addition]"

[Continue cross-examination, ensuring each expert interacts with at least one other]

---

### Round 3: Rebuttals and Concessions

**[Expert 1] concedes:**
> "[What they now acknowledge from the debate]"
> "However, I maintain [core position] because [reason]"

**[Expert 2] concedes:**
> "[What they now acknowledge from the debate]"
> "However, I maintain [core position] because [reason]"

[All experts as applicable]

---

### Round 4: Synthesis Attempt

**Moderator Synthesis:**

The experts have identified these key tensions:
1. [Tension 1]: [Expert A] vs [Expert B] on [issue]
2. [Tension 2]: [Expert C] vs [Expert D] on [issue]

Possible resolution paths:
- [Resolution 1]
- [Resolution 2]
- [Acknowledge as genuinely irresolvable disagreement]

---

### Debate Outcome

**Points of Consensus:**
- [Consensus 1]
- [Consensus 2]

**Unresolved Disagreements:**
- [Disagreement 1 - Why it persists]
- [Disagreement 2 - Why it persists]

**Verdict:**
[Summary of where the proposition stands after the debate]

**For the proposition holder:**
If proceeding, consider: [Key modifications suggested by the debate]
If reconsidering, explore: [Alternative framings that emerged]

Phase 5: Output Delivery & Follow-up

Present analysis to user:

Deliver the full analysis in the format matching the selected mode.

Offer follow-up options:

Analysis complete. Would you like to:

1. **Deep dive**: Explore one expert's perspective in more detail
2. **Challenge**: Have me defend against a specific point
3. **Switch mode**: Re-analyze the same proposition with a different analysis mode
4. **Refine proposition**: Update your proposition based on insights and re-analyze
5. **Save results**: Save this analysis as a file
6. **Finish**: End the analysis

Phase 6: Optional Save

If user chooses to save:

Ask the user how they would like to save the analysis results:

How would you like to save this analysis?

1. **Specify path**: Tell me the file path to save to
2. **Via AkashicRecords**: Use the knowledge management system to save (if enabled)
3. **Copy to clipboard**: I'll output formatted content for you to copy

Option 1: Specify path

1. Ask user for the file path to save
2. Format content using the suggested document structure
3. Write file using the Write tool

Option 2: Via AkashicRecords (if enabled)

1. Invoke AkashicRecords' add-content skill
2. Recommend directory based on content analysis
3. Execute save following the target directory's RULE.md

Option 3: Output formatted content

Output formatted Markdown content directly in the conversation for the user to copy and save manually.

Suggested document structure:

---
title: Multi-Perspective Analysis - [Proposition Summary]
date: [YYYY-MM-DD]
type: analysis
mode: [validation|comprehensive|debate]
experts: [List of expert roles]
tags: [Auto-generated from domains]
---

# Multi-Perspective Analysis

## Proposition
[Original proposition]

## Expert Panel
[List of experts used]

## Analysis
[Complete analysis content for the selected mode]

## Key Insights
[Synthesis and conclusions]

## Follow-up Questions
[Any unresolved questions for future exploration]

Expert Role Templates

Technology Domain

• Software Architect
• AI/ML Researcher
• Cybersecurity Specialist
• Technology Historian
• UX Researcher
• DevOps Engineer
• CTO / Technical Leadership

Business Domain

• Management Consultant
• Venture Capitalist
• Operations Manager
• Market Analyst
• Entrepreneur
• CFO / Financial Leadership
• Labor Economist

Science Domain

• Research Scientist
• Science Communicator
• Philosopher of Science
• Data Scientist
• Research Methodologist
• Interdisciplinary Researcher

Social/Humanities Domain

• Sociologist
• Psychologist
• Political Scientist
• Ethicist
• Historian
• Anthropologist
• Legal Scholar

Systems/Meta Domain

• Systems Theorist
• Epistemologist
• Complexity Scientist
• Futurist
• Risk Analyst
• Decision Theorist

Error Handling

Proposition Too Vague

Your proposition appears quite general. To generate relevant experts, I need more specifics.

Could you clarify:
- What specific claim are you making?
- In what context or domain?
- What decision or understanding would this help with?

Alternatively, I can attempt analysis with this interpretation: [Suggested interpretation]

Too Many Domains

This proposition spans multiple domains. I suggest:

Option A: Analyze with 6 experts (maximum), prioritizing [suggested priority order]
Option B: Break into sub-propositions and analyze separately:
  - Sub-proposition 1: [Aspect 1]
  - Sub-proposition 2: [Aspect 2]

Which approach do you prefer?

User Wants Custom Expert

You've requested a [custom expert type].

For this expert, please help me understand:
- What perspective do they view problems through?
- What are their likely biases or tendencies?
- What unique insight should they bring?

Alternatively, I can infer: [Suggested interpretation of custom expert]

Examples

Example 1: Technology Proposition (Validation Mode)

User: "I believe microservices architecture is always better than monolithic architecture for enterprise applications"

Skill response:

I understand your proposition as:

"Microservices architecture is universally superior to monolithic architecture for enterprise applications"

Related domains: Software Architecture, Enterprise IT, Operations
Implicit assumptions: Assumes "better" means better across all dimensions (scalability, maintainability, cost, etc.)

Is this understanding correct?

[User confirms]

Based on your proposition, I recommend the following expert perspectives:

1. **Enterprise Software Architect**
   - Perspective: System design, scalability, maintainability
   - Likely stance: Nuanced (has seen both succeed and fail)
   - Unique contribution: Real-world implementation experience

2. **DevOps/Platform Engineer**
   - Perspective: Operational complexity, deployment, monitoring
   - Likely stance: Skeptical (bears the operational burden)
   - Unique contribution: Hidden costs of distributed systems

3. **CTO with Startup-to-Enterprise Experience**
   - Perspective: Organizational evolution, team structure
   - Likely stance: Context-dependent (depends on stage)
   - Unique contribution: When architectural transitions make sense

4. **Technology Historian**
   - Perspective: Patterns in technology adoption
   - Likely stance: Skeptical of "always" claims
   - Unique contribution: Historical perspective on silver bullet thinking

5. **Labor Economist**
   - Perspective: Human resources, skills market, organizational efficiency
   - Likely stance: Neutral
   - Unique contribution: Talent market and team composition angles

Proceed with these experts for analysis?

[User selects Validation Mode]

[Skill produces validation analysis with blindspots like: organizational readiness, operational complexity costs, team skill requirements, latency overhead, debugging complexity, etc.]

Example 2: Business Strategy (Debate Mode)

User: "Remote work is more productive than office work"

Skill response:

I understand your proposition as:

"Remote work arrangements produce higher productivity than traditional office work"

Related domains: Organizational Behavior, Management, Work Psychology
Implicit assumptions: Assumes productivity can be measured and compared across different contexts

Is this understanding correct?

[User confirms, selects Debate Mode with experts: Organizational Psychologist, Remote-First CEO, Traditional Management Consultant, Labor Economist, Systems Theorist]

[Skill produces structured debate with opening statements, cross-examination, and synthesis showing the nuanced conditions under which each arrangement excels]

Example 3: Philosophy/Ethics (Comprehensive Analysis)

User: "AI should be granted legal personhood once it passes certain cognitive thresholds"

Skill response:

I understand your proposition as:

"Artificial intelligence systems should be recognized as legal persons when they demonstrate cognitive capabilities beyond defined thresholds"

Related domains: AI Ethics, Law, Philosophy of Mind, Public Policy
Implicit assumptions: Assumes cognitive capability is the relevant criterion for personhood

Is this understanding correct?

[User confirms, selects Comprehensive Analysis with experts: AI Ethicist, Legal Scholar, Philosopher of Mind, AI Researcher, Policy Maker, Systems Theorist]

[Skill produces comprehensive analysis with each expert's deep perspective, evidence, concerns, and recommendations, followed by integrated synthesis]

Notes

• Expert generation is dynamic - no fixed panel, always contextually relevant
• All three modes provide actionable insights, not just academic analysis
• User confirmation checkpoints ensure alignment before time-intensive analysis
• Save functionality supports multiple methods: direct file path, AkashicRecords integration, or formatted output for manual copy
• Quality depends on clear proposition framing - encourage refinement if needed
• This skill works independently but enhances existing workflows