/flux

Combination Engine

Purpose: Rules for dynamically combining frameworks and the Serendipity Injection mechanism. Read when: You are in the COMBINE phase or need to compose a multi-framework pipeline.

Contents

• Framework Compatibility Matrix
• Combination Rules
• Iterative Deepening Pipeline
• Serendipity Injection
• Example Combinations

Framework Compatibility Matrix

Compatibility scores (1-5): how well two frameworks amplify each other.

High-synergy pairs (score 5):

• First Principles + Assumption Reversal (decompose, then flip)
• First Principles + TRIZ (reduce to fundamentals, then resolve contradictions)
• Assumption Reversal + Devil's Advocate (surface, then attack)
• Assumption Reversal + Reframing (flip assumptions, then reframe the problem)
• Bisociation + Cross-Domain Analogy (two modes of domain crossing)
• Bisociation + Lateral Thinking (both break linear connections)
• Bisociation + Oblique Strategies (random stimulus amplifies bisociation)
• Lateral Thinking + Oblique Strategies (both inject randomness)
• Devil's Advocate + Multi-Agent Debate (structured opposition)

Combination Rules

Rule 1: Pillar Diversity

Always combine frameworks from at least 2 different pillars (CHALLENGE, COMBINE, SHIFT). Same-pillar-only combinations produce diminishing returns.

Good: First Principles (CHALLENGE) + SCAMPER (COMBINE) + Reframing (SHIFT) Avoid: First Principles + Assumption Reversal + Devil's Advocate (all CHALLENGE)

Rule 2: Synergy Threshold

Prefer combinations where all pairwise compatibility scores ≥ 3. Allow one pair below 3 only if it introduces essential pillar diversity.

Rule 3: Diminishing Returns

Maximum 5 frameworks per pipeline execution. Beyond 5, cognitive load outweighs additional insight.

Rule 4: Sequential Amplification

Order frameworks so each phase's output enriches the next:

CHALLENGE output (exposed assumptions)
    → feeds COMBINE input (assumptions become material for recombination)
        → feeds SHIFT input (combined ideas become material for perspective rotation)

Rule 5: Contradiction Preservation

When frameworks produce contradictory insights, preserve both. Contradictions are often the most valuable output — they reveal genuine tensions in the problem space.

Iterative Deepening Pipeline

Each phase transforms the problem representation:

Problem Statement (P₀)
    │
    ├─ CLASSIFY → Domain + Framework Set
    │
    ├─ CHALLENGE → P₁ (assumptions exposed, reversed)
    │     Input: P₀
    │     Output: Assumption Map + reversed problem variants
    │
    ├─ COMBINE → P₂ (cross-domain connections injected)
    │     Input: P₁ + Assumption Map
    │     Output: Bisociated concepts + SCAMPER variants + TRIZ solutions
    │
    ├─ SHIFT → P₃ (perspective rotated)
    │     Input: P₂ + all prior outputs
    │     Output: Reframed problem statements + lateral insights
    │
    └─ CRYSTALLIZE → Final Output
          Input: P₀, P₁, P₂, P₃ + all artifacts
          Output: 3-5 reframed problems + Insight Matrix + Blind Spot Report

Key principle: P₃ should be recognizably related to P₀ but offer genuinely different angles. If P₃ ≈ P₀, the pipeline hasn't produced sufficient transformation — escalate framework intensity.

Serendipity Injection

Concept

Deliberate introduction of random stimuli to break cognitive fixation. Applied during COMBINE and SHIFT phases.

Mechanism

1. Random Word Injection (Lateral Thinking → Random Entry)

   • Generate or select a random word unrelated to the problem.
   • Force 3 connections between the word and the problem.
   • Keep the most surprising connection.

2. Oblique Strategy Draw

   • Select 1-2 strategies randomly from the Oblique Strategies collection.
   • Apply as constraints or questions to current thinking.
   • Force at least one idea per strategy.

3. Domain Roulette (Bisociation)

   • Randomly select an unrelated domain.
   • Map the problem's structure onto that domain.
   • What solution would that domain suggest?

4. Provocation PO

   • Construct an intentionally absurd statement about the problem.
   • Extract the useful principle buried in the absurdity.

When to Inject

Serendipity Quality Check

After injection, verify:

• The stimulus produced at least one genuinely unexpected connection.
• The connection is meaningful, not just word-play.
• The insight would NOT have emerged from systematic analysis alone.

If all checks fail, inject again with a different mechanism.

Example Combinations

Complex Domain (User Churn Problem)

CLASSIFY: Complex (cause-effect retrospective only, multiple interacting factors)

CHALLENGE:
  - Assumption Reversal (20 assumptions about why users leave)
  - First Principles (what is the irreducible value we provide?)

COMBINE:
  - Bisociation: Churn × Restaurant loyalty → "regulars" concept
  - TRIZ: Contradiction "personalization requires data, data collection repels users"
  - Serendipity: Domain Roulette → "Gaming" → progression systems

SHIFT:
  - Reframing (Iceberg): Surface "users leave" → Pattern "they never truly arrived"
  - Oblique Strategy: "Honor thy error as a hidden intention" → churned users as product testers
  - Multi-Agent Debate: "Churn is healthy" vs "Churn is failure"

Result: 5 reframed problem statements

Complicated Domain (API Design Decision)

CLASSIFY: Complicated (expert analysis needed, multiple valid approaches)

CHALLENGE:
  - Assumption Reversal (10 assumptions about REST vs GraphQL)

COMBINE:
  - SCAMPER on existing API patterns
  - Cross-Domain Analogy: Library catalog systems

SHIFT:
  - Six Thinking Hats (quick rotation)

Result: 3 reframed problem statements

Chaotic Domain (Production Crisis Response)

CLASSIFY: Chaotic (unprecedented failure mode, no clear cause-effect)

CHALLENGE:
  - Quick Assumption List (5 assumptions about the failure)

COMBINE:
  - Random Entry: "bridge" → "what connects the failing components?"

SHIFT:
  - Oblique Strategy: "What would you do if given unlimited resources?"
  - PO Provocation: "PO: The system is working perfectly"

Result: 3 rapid reframings + immediate action hypotheses