Cross-domain patterns must share causal structure, not surface vocabulary
Validate cross-domain pattern candidates by verifying that the relational structure (not surface similarity) matches across domains—two patterns share structure when the causal relationships between elements are preserved even when the elements themselves differ completely.
Why This Is a Rule
Gentner's structure-mapping theory (1983) establishes that genuine analogies share relational structure, not surface features. "Technical debt accumulates like financial debt" is a valid cross-domain pattern because the causal structure is preserved: both involve deferred costs that compound over time. "Coding is like cooking" is a surface analogy — both involve creating something, but the causal relationships (ingredient interactions vs. logic dependencies) are fundamentally different.
Surface similarity feels insightful but misleads. Two situations that use similar vocabulary ("bottleneck" in manufacturing vs. "bottleneck" in psychology) seem structurally related but may have completely different causal mechanisms. Conversely, two situations with zero vocabulary overlap may share identical causal structure.
The validation test: strip the domain-specific elements from both patterns and compare the remaining relational skeleton. If the same causal relationships (A causes B, B limits C, removing C breaks D) appear in both domains, the cross-domain pattern is valid. If only the labels match, it's a surface analogy that will mislead if acted upon.
When This Fires
- When a pattern from one domain "reminds you of" a pattern in another domain
- During brainstorming when someone says "this is just like [other domain]"
- When building cross-domain links in your knowledge system (Prioritize cross-domain links over within-cluster links — weak ties generate better insights)
- Any time an analogy feels insightful and you want to verify it's genuine
Common Failure Mode
Accepting vocabulary overlap as pattern validation: "Both situations have a 'bottleneck,' so the same solution should work." But a manufacturing bottleneck (physical throughput constraint) requires different interventions than a cognitive bottleneck (attention capacity limitation). The word is the same; the causal structure is different.
The Protocol
When a cross-domain pattern candidate surfaces: (1) Describe the pattern in Domain A: what are the elements and what causal relationships connect them? (2) Describe the suspected pattern in Domain B with the same structure. (3) Compare: are the causal relationships preserved? Does A-causes-B in Domain 1 map to A'-causes-B' in Domain 2? (4) If causal structure is preserved → valid cross-domain pattern. Solutions that work in one domain may transfer. (5) If only surface features match → invalid analogy. Solutions will not transfer despite the superficial resemblance.