curiositech/tacit-expertise-elicitation

SKILL.md: Tacit Expertise Elicitation

Decision Points

Method Selection Decision Tree

Need to elicit expert knowledge?
├── Expert can articulate their process clearly
│   ├── YES → Suspect rehearsed account
│   │   └── Use situational re-evocation with specific difficult case
│   └── NO → Says "I just know" or "it's intuition"
│       └── Use MELI protocol with four-pass structure
│
├── Time/resource constraints?
│   ├── Single expert, 2-3 hours → MELI interview
│   ├── Multiple experts, short timeline → Cognitive shadowing
│   ├── Cross-cultural context → Community workshop (elicitive approach)
│   └── Long-term institutional knowledge → Cognitive task analysis
│
├── Knowledge transfer context?
│   ├── Within same domain/culture → Direct MELI
│   ├── Cross-domain transfer → Elicitive methodology first
│   ├── Retiring expert → Community-embedded MELI
│   └── Failed previous attempts → Diagnose structural barriers
│
└── Community dynamics?
    ├── Competitive/political tensions → Use community setting
    ├── High trust environment → Individual or small group
    └── Learning resistance → Focus on specific past moments

Interview Moment Decision Points

During elicitation session:
├── Expert gives smooth narrative → "Let's go back to that specific moment..."
├── Expert says "usually I..." → "In this particular case, what did you do?"
├── Expert rationalizes decision → "What did you notice first?"
├── Expert claims "best practice" → "What made you deviate from that?"
└── Expert shows uncertainty → "That uncertainty - what was driving it?"

Failure Modes

1. The Rehearsed Account Trap

Detection: Expert provides smooth, logical narrative with clear problem→analysis→solution arc Symptoms: No hesitation, no uncertainty mentioned, perfect hindsight clarity Fix: Interrupt with "Let's pause at that moment when you first noticed X. What exactly did you see?"

2. Real-Time Verbalization Error

Detection: You're asking experts to explain their thinking while performing Symptoms: Stilted performance, rationalized explanations, changed decision patterns Fix: Use post-hoc structured retrospective with sensory detail reconstruction

3. Credential Conflation

Detection: Selecting interviewees based on titles, certifications, or years of experience Symptoms: Knowledge that doesn't transfer, generic advice, inability to handle edge cases Fix: Look for evidence of deliberate practice and demonstrated adaptive expertise

4. Context Stripping

Detection: Extracting "universal principles" that should work everywhere Symptoms: Knowledge fails when applied in different contexts, local experts feel devalued Fix: Embed elicitation within target community, surface local knowledge first

5. Competitive Contamination

Detection: Formal CPD/learning settings producing curated success stories only Symptoms: Repeated organizational failures, knowledge hoarding, surface-level sharing Fix: Create psychologically safe structured protocols that make curation difficult

Worked Examples

Example 1: Senior Mediator Knowledge Capture

Setup: Law firm's top mediator retiring, needs to transfer 20 years of expertise to junior staff.

Novice approach would be: "Tell us your best practices" → Gets rehearsed principles about "active listening" and "finding common ground"

Expert elicitation process:

1. Identify specific difficult case from mediator's recent memory
2. First pass - situational reconstruction: "It's Tuesday morning, you walk into the conference room. What do you see? What's the energy in the room?"
3. Expert reveals: "The plaintiff's lawyer had his arms crossed, sitting back. I knew immediately this wasn't going to be a numbers negotiation"
4. Second pass - decision moments: "What made you decide to separate the parties right then instead of starting together?"
5. Expert reveals tacit pattern: "When I see that body language combo - crossed arms plus the particular way he was positioned relative to his client - it means he's performing for his client, not negotiating. Direct discussion would just entrench positions."
6. Third pass - constraints/alternatives: "What else could you have done? What stopped you?"
7. Fourth pass - transfer: Junior mediators practice recognizing this pattern in role-play

Key insight captured: Expert wasn't following "separate parties when tensions are high" - they were reading micro-signals of performative vs. genuine negotiation stance.

Example 2: Failed Knowledge Transfer Recovery

Setup: Manufacturing company's quality control training keeps failing - new inspectors miss defects that experienced ones catch easily.

Previous failed approach: Experienced inspectors created checklist of "what to look for" → New inspectors followed checklist but still missed critical issues

Elicitation process:

1. Community workshop setting - 3 expert inspectors, 2 trainees, facilitator
2. Specific case focus: Recent near-miss where trainee passed defective part
3. Reconstruct the moment: "You're looking at the part. Walk me through the first 10 seconds."
4. Expert reveals: "I don't actually go through the checklist sequentially. I scan for overall 'wrongness' first - something about the surface texture didn't look right"
5. Probe deeper: "What does 'wrongness' look like? What are your eyes doing?"
6. Tacit knowledge surfaced: Expert's scanning pattern follows stress lines in the material, accumulated over years of seeing failures
7. Transfer method: Create exercises where trainees practice pattern recognition on progressively subtle examples

Outcome: Training redesigned around pattern recognition rather than sequential checking.

Example 3: Cross-Cultural Expertise Transfer

Setup: International development organization wants to apply successful conflict resolution methods from Northern Ireland to post-conflict Africa.

Standard approach would be: Export the Northern Ireland framework directly → Predictable failure due to context mismatch

Elicitive approach:

1. Start with local expertise: Identify respected local practitioners who've successfully managed similar conflicts
2. MELI with local experts first: What are they actually doing? What constraints do they navigate?
3. Surface local knowledge: Community-embedded practices around reconciliation, authority structures, timing
4. Find bridge points: Where Northern Ireland experience resonates with local patterns
5. Hybrid development: Create methods that honor local knowledge while incorporating relevant external insights

Key principle: Local expertise gets elevated to equal status with imported expertise.

Quality Gates

Knowledge elicitation session is complete when:

• [ ] Expert has described at least one specific past moment in sensory detail
• [ ] At least one piece of tacit knowledge has been surfaced that wasn't in their initial account
• [ ] Decision constraints and alternatives have been explored, not just chosen actions
• [ ] Context factors that shaped the expert's options have been identified
• [ ] Pattern recognition or "intuitive" judgments have been unpacked into observable cues
• [ ] Knowledge has been tested with other community members present
• [ ] Transfer method has been designed that doesn't require identical context
• [ ] Political/structural barriers to knowledge adoption have been identified
• [ ] Expert confirms the elicited knowledge feels accurate to their actual practice
• [ ] Community members can identify how to apply insights in their own situations

NOT-FOR Boundaries

This skill should NOT be used for:

General interviewing → Use standard interview techniques instead

Performance evaluation → For assessment of individual performance, use performance management frameworks instead

Academic research → For theoretical knowledge generation, use research methodology instead

Crisis decision support → For real-time expert assistance, use decision support systems instead

Mediation practice itself → For conducting actual mediations, use conflict resolution skills instead

Organizational change management → For systemic organizational change, use change management frameworks instead

When expert knowledge is already well-documented → For codified knowledge transfer, use training design instead

When context cannot be preserved → For knowledge that must work across radically different contexts, use abstraction and generalization methods instead

This skill specifically targets the gap between what experts can say they do and what they actually do in practice - the tacit, context-dependent knowledge that resists explicit articulation but is essential for competent performance.