MECE Validation

Input: $ARGUMENTS

Purpose

Validate that a list achieves MECE quality:

• Mutually Exclusive (ME): No item belongs to multiple categories
• Collectively Exhaustive (CE): All possible items are covered

This skill identifies overlaps (ME violations) and gaps (CE violations).

MECE Spectrum

Perfect MECE is often impossible. This skill uses a soft MECE approach:

The Validation Process

Step 1: Confirm Structure

Identify what you’re validating:

VALIDATING: [list name/description]

STRUCTURE:
- Total items: [N]
- Categories: [list of categories if hierarchical]
- Dimensions covered: [if from space_enumeration]
- Target MECE level: [Perfect / Soft / Good Enough]

Step 2: Check Mutual Exclusivity (ME)

For each item, check: Does it fit in multiple categories?

MUTUAL EXCLUSIVITY CHECK:

| Item | Primary Category | Also Fits? | Overlap? |
|------|------------------|------------|----------|
| [Item 1] | [Cat A] | No | [x] ME |
| [Item 2] | [Cat B] | [Cat C] | [!] OVERLAP |
| [Item 3] | [Cat A] | No | [x] ME |
...

OVERLAPS FOUND: [N]
OVERLAP RATE: [N / Total] = [X%]

For each overlap, decide:

1. Assign to primary: Pick one category, note secondary
2. Split the item: Create more specific sub-items
3. Merge categories: If overlap is systematic
4. Accept overlap: If categories are perspectives, not buckets

Step 3: Check Collective Exhaustiveness (CE)

3A: Dimension Coverage Check

If dimensions are known (from /dimension_discovery):

DIMENSION COVERAGE:

| Dimension | Values | Items Found | Missing? |
|-----------|--------|-------------|----------|
| [Dim 1] | [v1, v2, v3] | [3, 2, 4] | No |
| [Dim 2] | [v1, v2] | [5, 0] | YES: v2 |
...

GAPS FOUND: [N]
GAP RATE: [Missing combinations / Total possible] = [X%]

3B: Edge Case Check

Check for items that might be missing:

EDGE CASE PROBES:

| Probe | Found in List? | If No, Add? |
|-------|----------------|-------------|
| Extreme cases (max/min) | [Y/N] | [action] |
| Boundary cases | [Y/N] | [action] |
| Negative cases (what it's NOT) | [Y/N] | [action] |
| Historical cases (past examples) | [Y/N] | [action] |
| Future cases (emerging trends) | [Y/N] | [action] |
| Edge stakeholders | [Y/N] | [action] |

3C: “What’s Missing?” Brainstorm

Ask explicitly: What could be on this list but isn’t?

WHAT'S MISSING BRAINSTORM:

Domain expert would add:
- [potential missing item 1]
- [potential missing item 2]

Contrarian would add:
- [unconventional item that belongs]

Adjacent domain has:
- [item from related field that applies]

Step 4: Calculate MECE Score

MECE SCORE:

Mutual Exclusivity:
- Overlaps: [N] / [Total] = [X%]
- ME Score: [100 - X]%

Collective Exhaustiveness:
- Gaps identified: [N]
- Estimated coverage: [Y%]
- CE Score: [Y]%

OVERALL MECE: (ME Score + CE Score) / 2 = [Z%]

ASSESSMENT:
- [ ] Perfect MECE (>95%)
- [x] Soft MECE (80-95%)
- [ ] Good Enough (60-80%)
- [ ] Needs Work (<60%)

Step 5: Generate Recommendations

Based on findings:

MECE VALIDATION RESULTS: [list name]

STATUS: [Perfect MECE / Soft MECE / Good Enough / Needs Work]

OVERLAPS TO RESOLVE:
1. [Item X] overlaps [Cat A] and [Cat B]
   -> Recommendation: [assign/split/merge/accept]

GAPS TO FILL:
1. [Dimension Y, Value Z] has no items
   -> Recommendation: [add specific items]

2. [Edge case] not covered
   -> Recommendation: [add item]

ACTIONS:
- [ ] [Specific action 1]
- [ ] [Specific action 2]

Example: Validating “Types of Machine Learning”

Input List

1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning
4. Deep Learning
5. Neural Networks
6. Semi-supervised Learning

Step 1: Structure

• Total items: 6
• Target: Soft MECE

Step 2: ME Check

Overlaps: 2 (Deep Learning, Neural Networks) Analysis: List mixes two dimensions - paradigm and architecture

Step 3: CE Check

Dimension coverage:

• Paradigms: Supervised, Unsupervised, Reinforcement, Semi-supervised [x]
• Missing: Self-supervised learning

Edge cases:

• Transfer learning? Not covered
• Online learning? Not covered

Step 4: Score

• ME: 4/6 clean = 67%
• CE: Missing ~3 items from ~10 possible = 70%
• Overall: 68% = Good Enough (Needs Work)

Step 5: Recommendations

STATUS: Good Enough - Needs Work

OVERLAPS TO RESOLVE:
1. "Deep Learning" and "Neural Networks" are architecture, not paradigm
   -> Split list into: A) Learning Paradigms, B) Architectures
   OR -> Remove Deep Learning/Neural Networks (different dimension)

GAPS TO FILL:
1. Self-supervised learning (emerging paradigm)
2. Transfer learning (important paradigm)
3. Online/incremental learning (paradigm)

RECOMMENDED REVISED LIST:
Learning Paradigms:
- Supervised
- Unsupervised
- Semi-supervised
- Self-supervised
- Reinforcement
- Transfer learning
- Online learning

Quick MECE Check (Abbreviated)

For fast validation:

QUICK MECE: [list name]

[ ] Any item fits 2+ categories? -> If yes, overlap issue
[ ] Any obvious gaps? -> If yes, CE issue
[ ] Would domain expert add anything? -> If yes, gap
[ ] Would domain expert remove anything? -> If yes, scope issue

QUICK VERDICT: [MECE / Mostly MECE / Needs Work]

Quality Checklist

Before completing:

• Structure confirmed
• All items checked for overlaps
• Dimension coverage verified
• Edge cases probed
• “What’s missing” brainstormed
• MECE score calculated
• Specific recommendations provided

Integration

Use with:

• /dimension_discovery -> Identify dimensions first
• /space_enumeration -> Generate the list
• /mece_validation -> Validate the list (this skill)