微信扫一扫Analogical Reasoning
Transfer structured knowledge across domains. The logic of pattern recognition and adaptation.
Type Signature
Analogical : Source → StructuralMap → Target → Adaptation
Where:
Source : PriorExperience × Relevance → SourceDomain
StructuralMap : SourceDomain → (Objects × Relations × Constraints)
Target : StructuralMap × NewContext → MappedStructure
Adaptation : MappedStructure × ContextDifferences → AdaptedSolution
When to Use
Use analogical when:
- Entering new market with experience in similar markets
- Building new product with experience in similar products
- Facing novel situation with structural similarity to past cases
- Need to transfer playbooks across contexts
- "This is like..." patterns in thinking
Don't use when:
- Cause-effect chain is known → Use Causal
- Need to explain observation → Use Abductive
- Competing positions to resolve → Use Dialectical
Four-Stage Process
Stage 1: Source Retrieval
Purpose: Identify relevant prior experience with documented outcomes.
Source Selection Criteria:
| Criterion | Question | Weight | |-----------|----------|--------| | Structural similarity | Same type of problem/situation? | 0.35 | | Outcome documented | Do we know what happened? | 0.25 | | Recency | How recent is the experience? | 0.15 | | Success level | Did the approach work? | 0.15 | | Context overlap | Similar constraints/resources? | 0.10 |
Source Retrieval Process:
retrieval:
query: "Entering B2B marketplace vertical"
candidates:
- source: "Shopify DTC launch (2024)"
similarity: 0.75
outcome: "Validated in 6 months, $200K ARR"
success: high
- source: "Fashion brand pilot (2023)"
similarity: 0.60
outcome: "Slow start, pivoted twice"
success: medium
- source: "Enterprise SDK launch (2024)"
similarity: 0.50
outcome: "$400K first deal, strong pipeline"
success: high
selected: "Shopify DTC launch"
reason: "Highest structural similarity (platform integration,
API-first, self-serve onboarding)"
Output:
source:
case: "Shopify DTC launch"
domain: "E-commerce platform integration"
timeframe: "Q1-Q2 2024"
outcome:
result: "success"
metrics: "$200K ARR, 50 merchants, 6-month validation"
key_factors:
- "Strong app store presence"
- "Self-serve onboarding"
- "Integration-first positioning"
documented_in: "threads/operations/shopify-dtc-launch/"
Stage 2: Structural Mapping
Purpose: Extract transferable structure from source domain.
Mapping Components:
| Component | Source Example | Abstracted | |-----------|----------------|------------| | Objects | Shopify merchants | Platform users | | Relations | Merchant → App → Customer | User → Integration → End-user | | Constraints | App store rules | Platform policies | | Mechanisms | App store discovery → trial → purchase | Discovery → trial → convert | | Success factors | Reviews, featured placement | Social proof, visibility |
Structural Map:
structure:
objects:
- User: "Entity adopting our solution"
- Platform: "Ecosystem we integrate with"
- EndUser: "Final beneficiary of solution"
- Solution: "Our product/integration"
relations:
- Platform ⊃ Marketplace: "Platform has discovery mechanism"
- User → Solution: "User adopts solution"
- Solution → EndUser: "Solution serves end users"
- EndUser feedback → User: "Value demonstration"
mechanisms:
acquisition:
- "Platform marketplace discovery"
- "Peer recommendations"
- "Content marketing to users"
activation:
- "Self-serve trial"
- "Quick time-to-value"
- "Integration simplicity"
retention:
- "Embedded in workflow"
- "Switching cost creation"
- "Continuous value delivery"
constraints:
- "Platform approval required"
- "Platform policies must be followed"
- "Revenue share with platform"
success_factors:
- "Marketplace ranking/visibility"
- "User reviews/ratings"
- "Platform relationship quality"
Stage 3: Target Application
Purpose: Map structure to new context, identifying what transfers and what doesn't.
Target Context:
target:
domain: "B2B marketplace integration"
platform: "Faire wholesale marketplace"
user: "Wholesale brands"
end_user: "Retailers"
goal: "Return reduction for wholesale fashion"
Mapping Execution:
mapping:
objects:
Platform: "Shopify" → "Faire"
User: "DTC merchant" → "Wholesale brand"
EndUser: "Consumer" → "Retailer"
Solution: "Fit recommendation app" → "Wholesale sizing tool"
relations:
preserved:
- "Platform marketplace discovery" (Faire has app marketplace)
- "User adopts solution" (brands install integrations)
- "Value to end user" (retailers get better sizing)
modified:
- "Self-serve trial" → "Account executive assisted"
reason: "B2B decision process differs"
- "Individual purchase" → "Contract-based"
reason: "Wholesale pricing models"
broken:
- "App store reviews drive adoption"
reason: "Faire marketplace less review-driven"
replacement: "Case studies and referrals"
mechanisms:
acquisition:
transfers: "Platform marketplace presence"
adapts: "Content marketing → Trade show presence"
new: "Wholesale buyer referral program"
activation:
transfers: "Integration simplicity"
adapts: "Self-serve → Assisted onboarding"
new: "Pilot with single retail partner"
retention:
transfers: "Embedded in workflow"
transfers: "Value demonstration"
adapts: "Individual metrics → Fleet metrics"
Stage 4: Adaptation
Purpose: Produce concrete plan adjusted for context differences.
Context Differences Analysis:
differences:
critical:
- name: "Decision process"
source: "Individual merchant, fast"
target: "Buying committee, slow"
adaptation: "Add sales support, longer cycle expectations"
- name: "Value demonstration"
source: "Per-order metrics visible"
target: "Aggregate across retailers"
adaptation: "Build analytics dashboard for brands"
moderate:
- name: "Pricing model"
source: "Per-store subscription"
target: "Volume-based or percentage"
adaptation: "Explore usage-based pricing"
minor:
- name: "Technical integration"
source: "Shopify API"
target: "Faire API"
adaptation: "Standard integration work"
Adapted Solution:
adaptation:
strategy: "Platform-assisted B2B wholesale launch"
what_transfers:
- "Integration-first positioning"
- "Platform relationship investment"
- "Quick time-to-value focus"
- "Embedded workflow stickiness"
what_adapts:
- "Self-serve → Assisted onboarding with demo"
- "App store discovery → Trade shows + referrals"
- "Individual reviews → Case studies"
- "Per-order metrics → Brand-level analytics"
what's_new:
- "Sales motion for wholesale buyers"
- "Multi-retailer aggregation features"
- "B2B pricing model (volume-based)"
execution_plan:
phase_1: "Platform partnership + 3 pilot brands"
phase_2: "Case study development + trade show presence"
phase_3: "Scale via referrals + platform promotion"
expected_timeline: "9-12 months (vs 6 months for DTC)"
reason: "B2B sales cycle longer, relationship-building required"
confidence: 0.70
uncertainty:
- "Faire marketplace dynamics unknown"
- "Wholesale brand decision process may vary"
- "Volume-based pricing acceptance unclear"
Quality Gates
| Gate | Requirement | Failure Action | |------|-------------|----------------| | Source quality | Documented outcome with metrics | Find better source | | Structural clarity | ≥3 objects, ≥3 relations explicit | Complete mapping | | Mapping coverage | All source elements mapped or marked broken | Complete mapping | | Adaptation specificity | Concrete actions, not abstract | Add specificity | | Confidence threshold | ≥0.6 confidence | Flag high uncertainty |
Common Failure Modes
| Failure | Symptom | Fix | |---------|---------|-----| | Surface similarity | Mapped by superficial features, not structure | Focus on relations, not objects | | Over-transfer | Assume everything applies | Explicitly check each element | | Under-adaptation | Copy-paste without adjustment | Force context difference analysis | | Single source | Only one analogy considered | Retrieve multiple candidates |
Output Contract
analogical_output:
source:
case: string
domain: string
outcome: {result: string, metrics: string}
thread_ref: optional<string>
mapping:
objects: {source_name: target_name}
relations:
preserved: [string]
modified: [{from: string, to: string, reason: string}]
broken: [{relation: string, reason: string, replacement: string}]
adaptation:
transfers: [string] # What applies directly
adapts: [string] # What needs modification
new: [string] # What's genuinely new
plan:
phases: [{name: string, actions: [string]}]
timeline: string
milestones: [string]
confidence: float # 0.0-1.0
uncertainty: [string]
next:
suggested_mode: ReasoningMode # Usually causal
canvas_refs: [string] # Assumptions being tested
trace:
sources_considered: int
mapping_coverage: float # % of source elements mapped
duration_ms: int
Example Execution
Context: "Expand to home goods vertical (currently in fashion)"
Stage 1 - Source Retrieval:
Selected: Fashion DTC success (highest similarity)
Alternatives considered:
- Beauty vertical (rejected: different return dynamics)
- B2B wholesale (rejected: different buyer)
Stage 2 - Structural Mapping:
Objects: Fashion brand → Home goods brand
Relations:
- Fit concern → Dimension/space concern
- Style matching → Aesthetic matching
- Return reason: fit → Return reason: scale/compatibility
Mechanisms:
- Visual AI → Transfer (image analysis)
- Size recommendation → Adapt (dimension recommendation)
- Color matching → Transfer (palette matching)
Stage 3 - Target Application:
Preserved: Visual AI core, recommendation engine, integration model
Modified: Fit algorithm → Dimension/space algorithm
Broken: Body measurement input → Room/space measurement input
Stage 4 - Adaptation:
Plan:
Phase 1: Partner with 2 home goods DTC brands (furniture focus)
Phase 2: Adapt algorithm for dimension-based recommendations
Phase 3: Develop room visualization feature (new capability)
Timeline: 4-6 months (faster than fashion - simpler measurements)
Confidence: 0.75
Key uncertainty: Room visualization technical complexity