Skip to content

Analogy Mapping Canvas

A structured worksheet — instantiates Design-Principle Extraction and Reapplication

Lays a source and a target side by side and aligns their relational structure — not their surface features — so an extracted principle can be re-expressed in the target's own materials.

The Analogy Mapping Canvas is a two-column worksheet that puts the source and the receiving context face to face and forces every element of the source to name its counterpart in the target — or be marked no analog. Its defining move is that it maps the relations between parts, not the parts' surface attributes: what matters for transfer is which element plays which role and how the roles connect, so the same logic can be rebuilt out of entirely different materials. It is the mechanism that turns "we should do something like they did" into an explicit correspondence table and a set of claims about what must change — and, just as usefully, into a visible list of source elements that have no target counterpart at all.

Example

A platform team admires how wildfire crews run on the Incident Command System — one Incident Commander with clear authority, a bounded span of control, and standard roles that snap together even when strangers meet at a fire. They want their 2 a.m. outage response to feel that calm. On the canvas, the left column holds the ICS elements (Incident Commander, Operations Section, span of control ≈5–7, a single staging area); the right column asks for each one's counterpart in an on-call rotation. The rich part is the relation lines: "the IC holds decision authority so responders don't debate mid-fire" maps cleanly to "the incident lead owns the sev-call so engineers stop arguing in the channel." A physical staging area, though, has no analog — it gets marked no-transfer rather than cargo-culted into some pointless ritual.

What drops out is not a copy but a translation plus a hypothesis: "if one named incident lead can page anyone and declare severity, coordination time falls" — a claim precise enough to be wrong, attached to the specific relation that carries the principle.

How it works

What distinguishes the canvas from generic "compare and adapt" is three disciplines: it demands relational correspondence (align how X affects Y, not X looks like Y'); it treats an unmapped source element as a finding, not a gap to paper over, because a missing counterpart is exactly where a naïve copy would fail; and it emits one adaptation hypothesis per mapped relation, so the translation ships as testable claims rather than a vibe. The output is the filled correspondence table plus that hypothesis list — handed to a prototype to actually test.

Tuning parameters

  • Mapping grain — whether you align single elements, relations, or whole subsystems. Coarser grain maps faster but hides where the analogy snaps; finer grain catches breaks at the cost of effort.
  • No-analog strictness — how readily a source element is allowed to sit unmapped. Loose marking forces false counterparts (cargo cult); strict marking surfaces honest gaps but leaves the picture ragged.
  • Attribute-vs-relation weighting — how much a shared appearance is allowed to count as a correspondence. Leaning toward relations is safer for transfer; leaning toward attributes is faster but where false analogies live.
  • Cardinality — whether one-to-one, many-to-one, or one-to-many mappings are permitted. Allowing many-to-one captures principles that consolidate roles, but blurs accountability in the target.

When it helps, and when it misleads

Its strength is that it makes the break points of an analogy visible: the empty right-hand cells are the argument against a lazy copy, and the relation lines are what let a principle survive a change of materials. It is the natural front end to translation because it converts admiration into a checkable map.

Its failure mode is the false analogy — mapping things that merely look alike but play different roles, so the "translation" imports surface and drops function. The classic misuse is running the canvas backwards: deciding to copy first, then filling cells to dress the decision as reasoning. The discipline that guards against both is Gentner's rule of thumb — a good analogy transfers a system of relations, not a bag of matched attributes[1] — so a correspondence only counts if the relation it names actually does work in the target.

How it implements the components

  • target_context_translation_map — the filled two-column table is the translation map: it states, element by element, what in the target plays each source role and what must change for the principle to function there.
  • adaptation_hypothesis — each mapped relation is emitted as a testable claim about the change the target requires, ready to be prototyped.

It does not extract or record the principle itself (that's the Design Principle Card), compare several sources (that's the Benchmark Deconstruction Grid), or run the actual test (that's the Transfer Prototype Experiment).

References

[1] Dedre Gentner's structure-mapping theory holds that a strong analogy aligns the relational structure of two domains — the system of relations among objects — rather than the objects' surface attributes. It is the standard reason a canvas that maps roles and relations transfers, while one that maps appearances produces a false analogy. ↩