Psychologist Robin Hogarth divided learning environments into two types. The labels sound cute; the distinction explains when experience makes you better and when it makes you confidently wrong.
Kind environments
In a kind environment, patterns repeat, feedback is rapid and accurate, and the rules stay stable. Chess, golf, classical piano performance, firefighting in familiar building types, and surgical procedures you have done a thousand times all qualify. You can chunk recurring patterns (chunking) and improve through deliberate practice. Gary Klein's naturalistic decision making found that experts in these domains often know their move in seconds — pattern recognition, not slow analysis.
Kind environments are the implicit model behind the Tiger Woods path, the Polgar chess experiment, and the ten-thousand-hours rule.
Wicked environments
In a wicked environment, rules are incomplete or shifting, patterns may not repeat or may be hidden, and feedback is delayed, inaccurate, or both. Worse: experience can teach exactly the wrong lesson. Hogarth's typhoid doctor felt tongues and "confirmed" diagnoses that were really cross-contamination. Expert accountants given a new tax law did worse than novices. Bridge experts struggled when play order changed.
Daniel Kahneman's heuristics-and-biases work and Klein's NDM program looked like a paradox — does experience help or not? Their joint paper resolved it: it depends entirely on the domain. Narrow experience helps in kind worlds; in wicked worlds it often breeds confidence without skill.
Epstein calls most of real life "Martian tennis": you see something like a game, but the rules were never explained and may change without notice. Most professional work — medicine outside routine procedures, policy, investing, organizational leadership, scientific discovery — sits on the wicked end of the spectrum.
What to do in wicked worlds
- Treat narrow expertise as necessary but insufficient; seek distant analogies and the outside view.
- Vary training deliberately to avoid cognitive-entrenchment; keep one foot outside your world.
- Prefer test-and-learn and match-quality exploration over locking onto one path before you know the domain.
- Learn slowly with desirable-difficulties so knowledge stays flexible, not performative.
Implications
- Early hyperspecialization pays off mainly in kind domains with stable rules.
- In wicked domains, range, analogies, and evaluating before choosing beat instinctive pattern match.
- cognitive-entrenchment and man-with-a-hammer-syndrome are what happen when kind-world habits get exported to wicked problems.
- AI is increasingly savant-like: crushing kind games, struggling in open worlds where humans still add strategy (range-why-generalists-triumph).
Sources
- range-why-generalists-triumph
- thinking-fast-and-slow — Kahneman–Klein joint paper resolving the expert-intuition paradox; kind vs wicked as the boundary condition