Deep Structural Analogy
Solve new problems by mapping them to distant but structurally similar ones
Deep structural analogy is the practice of solving problems by identifying structural similarities between the current challenge and situations from entirely different domains. Rather than relying on surface-level similarities or expertise in the specific domain of the problem, this approach draws on diverse knowledge to find deep parallels that illuminate solutions.
Johannes Kepler revolutionized astronomy by analogizing the movement of planets to the behavior of light, boat oars, magnets, and other physical phenomena from completely different fields. He found solutions that specialists in astronomy had missed for centuries because he was willing to import ideas from distant domains. Research shows that the further afield an analogy comes from, the more likely it is to produce a creative breakthrough, but most people default to the nearest and most obvious comparisons.
The 'outside view,' a concept from Kahneman, is a specific application: instead of focusing on the unique internal details of a situation (the 'inside view'), you identify a reference class of structurally similar situations from different contexts and use their base rates to inform your prediction. Studies show this approach dramatically improves forecasting accuracy for everything from infrastructure projects to movie revenues.
- Surface similarities are seductive but misleading; deep structural similarities are harder to spot but far more useful
- The further afield a useful analogy comes from, the more creative and valuable the solution it produces
- The 'inside view' (focusing on unique details of the current case) systematically produces overconfident and inaccurate predictions
- The 'outside view' (comparing to a reference class of structurally similar cases) dramatically improves accuracy
- Experts tend to think within their domain; breakthroughs often come from importing ideas across domains
- Identify the deep structure of your problemStrip away surface details and identify the underlying abstract structure. What type of problem is this at its core? Is it a convergence problem, a resource allocation problem, a timing problem, a coordination problem?Pro tipIn the classic 'radiation problem,' a tumor must be destroyed without damaging surrounding tissue. The deep structure is: how to concentrate force on a target through resistant material. A military analogy (converging small forces from multiple directions) provides the solution.
- Generate analogies from distant domainsBrainstorm situations from completely different fields that share the same deep structure. The more different the surface features, the better. Push past the first few obvious analogies to find distant ones.Pro tipResearch showed that only 10% of college students spontaneously used an analogy from a different domain to solve a problem, but 80% could use one when it was pointed out. The analogies are available; you just need to look for them.WarningDo not stop at the first analogy you find. Generate multiple analogies and compare what they suggest.
- Apply the outside viewBefore diving into the internal details of your situation, find a reference class of broadly similar projects, decisions, or events and examine their base rates. How long did similar projects take? What percentage succeeded? What were the common failure modes?Pro tipNinety percent of major infrastructure projects come in over budget. If you are planning one, the outside view says you should plan for overruns regardless of how unique your project feels.WarningThe inside view feels more informative because it is richer in detail. That richness is precisely what makes it misleading. More detail leads to more confidence, not more accuracy.
- Integrate inside and outside viewsStart with the outside view as your baseline, then adjust modestly based on truly unique internal features. Do not let the internal details overwhelm the base rate.Pro tipNetflix improved its recommendation algorithm by combining a system that analyzed internal details of movies with one that used broad analogies to users' overall taste patterns.
Johannes Kepler analogized planetary motion to the behavior of light (intensity decreasing with distance), boat oars (force and motion), and magnets (attraction at a distance). These analogies from completely different physical domains allowed him to conceive of a force emanating from the sun that moved the planets, an idea no astronomer had considered.
Researchers gave investors detailed information about actual start-ups and asked them to predict success. Investors who spontaneously drew analogies to ventures in different industries made more accurate predictions than those who only analyzed the specific venture's internal details.
Epstein traces the concept to Kepler, who used analogies from optics, magnetism, and rowing to develop the first physics-based model of the solar system. The psychological research comes from Dedre Gentner at Northwestern, who showed that experts categorize problems by deep structural features while novices categorize by surface features. Kahneman's 'outside view' emerged from his own experience: he and a team were writing a curriculum and estimated it would take two years, ignoring the base rate that similar projects took seven to ten years. It took eight. The pattern appears across domains: venture capitalists who used broad analogies made better investment predictions than those who focused on deep analysis of the specific case.