LLM Book Knowledgebase

Analogical Thinking and the Outside View

8 min read

Analogical Thinking and the Outside View

Analogical thinking and the “outside view” are two related cognitive tools that address a common failure mode: over-relying on the specific details of the current problem — the inside view — at the expense of what can be learned from structurally similar cases in different domains. Both David Epstein (Range, 2019) and Annie Duke (Thinking in Bets, 2018) independently arrive at the same prescription for this failure from different directions: use reference classes, seek structural parallels, and resist the temptation to treat every situation as uniquely incomparable to anything else.

The Inside View: The Natural Default

The inside view is Daniel Kahneman’s term for the approach we naturally take when planning, estimating, or deciding: we focus on the specific features of the case in front of us, consider the unique circumstances and people involved, and build a projection from those particulars.

The inside view feels rigorous. We are, after all, paying close attention to what is actually happening. But research consistently shows it produces systematic errors:

“We take the inside view when we make judgments based narrowly on the details of a particular project that are right in front of us.” — David Epstein, Range

The planning fallacy — the near-universal tendency to underestimate how long projects take and how much they cost — is the inside view in action. You think about your specific project: your team, your plan, your particular advantages. You do not think about the base rate of similar projects, most of which ran over time and budget even when their planners felt confident.

The Outside View: Using Reference Classes

The outside view corrects the inside view by asking: what actually happened to a representative sample of situations structurally similar to this one?

“Our natural inclination to take the inside view can be defeated by following analogies to the ‘outside view.’ The outside view probes for deep structural similarities to the current problem in different ones. The outside view is deeply counterintuitive because it requires a decision maker to ignore unique surface features of the current project, on which they are the expert, and instead look outside for structurally similar analogies.” — Epstein, Range

A reference class is the set of similar cases you use to ground the outside view. It might be: all software projects of similar scope and team size. Or all regulatory approval processes of comparable complexity. Or all firsttime novel publications by debut authors. The reference class discipline forces you to ask whether your situation is really as unique as the inside view makes it feel.

“Using a full ‘reference class’ of analogies — the pillar of the outside view — was immensely more accurate.” — Epstein, Range

The Netflix recommendation algorithm is Epstein’s cleanest example: rather than decomposing each user’s preferences into sophisticated taste profiles (inside view), Netflix found it more accurate to simply find users with similar viewing histories (outside view). The complexity of prediction is handled by the reference class, not by the analyst’s theory.

Deep Analogical Thinking

Epstein’s treatment goes beyond the reference class into a richer form of creative problem solving: deep analogical thinking.

“Deep analogical thinking is the practice of recognizing conceptual similarities in multiple domains or scenarios that may seem to have little in common on the surface.”

The key word is deep. Surface analogies — comparing your startup to another startup, your medical situation to a similar medical situation — are useful but limited. Deep analogical thinking crosses domain boundaries to find structural parallels:

  • A physicist recognizes that the mathematical structure of a fluid dynamics problem is identical to an electrical circuit problem
  • A management consultant recognizes that the coordination failure in a healthcare system is structurally identical to a game theory problem in economics
  • A scientist looking at an anomalous result recognizes that a similar anomaly in a different field led to a major discovery

The experimental evidence for the power of deep analogy is striking. In a classic study, participants were asked to solve the “radiation problem” (destroy a tumor with rays without damaging surrounding tissue):

“Gift of a single analogy from a different domain tripled the proportion of solvers who got the radiation problem. Two analogies from disparate domains gave an even bigger boost.” — Epstein, Range

The fortress story — an army that had to divide into small groups approaching from multiple directions, because direct assault was blocked — is structurally identical to the radiation problem (use multiple weaker rays converging on the tumor from different directions). The domains are completely different; the deep structure is the same.

The Power of More, Distant Analogies

Epstein’s research finding is that the optimal analogical strategy is to use more analogies from more distant domains, not one analogy from the most similar domain. The intuition to find the closest possible analogy — the business most similar to yours, the experiment most similar to yours — produces exactly the wrong result. Distant analogies provide more independent information and are less likely to share the same blind spots.

Breadth Enables Better Analogies

The practical implication for personal development: people who have worked across more domains have a larger library of structural patterns to draw from. When they encounter a new problem, they have more potential analogies to test:

“‘I especially love analogies,’ he wrote, ‘my most faithful masters, acquainted with all the secrets of nature. One should make great use of them.‘” — Johannes Kepler, quoted in Range

Generalists and polymaths are not merely people with interesting cocktail party knowledge. They are people with more analogical resources, which translates into better problem-solving in genuinely novel situations. The domain specialist who has only ever worked in one field is limited by the Einstellung effect: they can only see their familiar solution paths.

This is the deeper argument for breadth in Range: it’s not that generalists are better at any specific task than specialists. It’s that they have more tools, and in a wicked world where problems don’t announce their type in advance, having more tools means better solutions.

The Outside View in Decision-Making Under Uncertainty

Annie Duke’s Thinking in Bets develops the outside view argument in the context of decision-making rather than creativity. The connection is direct: resulting (judging decision quality by outcome) is the inside view applied to learning. The specific outcome of the specific decision we made is the inside view data point. The outside view asks: across all decisions with a similar probability structure, what is the expected distribution of outcomes?

“Any prediction that is not 0% or 100% can’t be wrong solely because the most likely future doesn’t unfold.” — Annie Duke, Thinking in Bets

If you made a decision that had a 70% chance of success and it failed, the inside view says “I was wrong.” The outside view says “I was right on the merits; this was the 30% outcome.” The distinction matters enormously for learning: the inside view causes you to update your process based on one data point. The outside view keeps you calibrated across the full probability distribution.

Duke’s “Wanna bet?” trigger is a device for forcing the outside view in real time: before committing to a belief, ask yourself whether you’d be willing to stake something on it. This naturally prompts you to consider the reference class of situations similar to this one.

Reference Classes Are Not Infallible

The outside view has a failure mode of its own: choosing the wrong reference class. If your project is structurally unlike the projects in your reference class — if you really are doing something genuinely unprecedented — then the base rates from that class don’t apply. The outside view is most powerful when combined with careful analysis of whether the structural parallels actually hold. Kahneman’s recommendation is to begin with the outside view and then adjust for specific features of the current case that genuinely distinguish it — not the reverse.

Superforecasters as Outside-View Practitioners

Epstein’s treatment of Tetlock’s superforecasters — the ordinary people who dramatically outperform domain experts at prediction — shows that outside-view thinking is one of their defining characteristics. Superforecasters:

  • Proactively seek analogies from different domains
  • Actively resist the temptation to treat the current situation as unique
  • Maintain and update reference classes based on new evidence
  • Hold beliefs tentatively and revise them when new data arrives

This is Epstein’s fox in practice. The fox’s advantage is precisely the breadth of analogical resources that narrow experts lack.

Backlinks