Intuition and Expertise
Intuition is “the way we translate our experience into action” [1] — a skill that can be built, applied and safeguarded, not a gift. Gary Klein’s Recognition-Primed Decision (RPD) model describes how experts decide under pressure.
1. What RPD claims
Klein built RPD from studies of fireground commanders, later military and police [2]. The claim is not that experts choose well among options — they usually generate none. A recognised pattern produces one candidate action, run forward mentally; if the simulation exposes a problem, it is modified or the next candidate considered. Klein ties this to Simon’s satisficing: options are evaluated one at a time and the first workable one taken — hence the expert’s good-enough first considered action.
The chain: cues → patterns → action scripts → mental simulation, all driven by mental models [1].
2. Three variations, and the safeguard
RPD is not one chain but three.
- Simple match — the situation is recognised and the action taken.
- Diagnose the situation — cues do not match cleanly, so the decider gathers more, feature-matches, or builds a story.
- Evaluate a course of action — the recognised action is simulated, then modified or rejected.
Expectancies are a first-class element alongside cues, goals and actions, making the model self-correcting: a violated expectancy is how an expert discovers the interpretation was wrong. Without them RPD is merely fast.
Example. A tech lead sees the nightly build fail on three unrelated tests. The pattern shared fixture state fires, with an expectancy: those three should pass in isolation. They do not, so she drops it within a minute for a clock-dependent assertion — variation 2.
3. How often experts work this way
Across more than six hundred decision points [2], RPD accounted for 46% to 96% of decisions by domain and expertise: urban fireground commanders 80% of 156 points, design engineers 60%, AEGIS commanders 95%, tank-platoon trainees 42%.
4. When to recognise, when to compare
Recognition is fast and allows action under uncertainty; comparison buys justification. Klein’s boundary conditions [2]:
| Recognition works when… | Compare options when… |
|---|---|
| Time pressure is high | The choice must be justified to others |
| The decider is experienced in this domain | Competing interests must be reconciled |
| Conditions are dynamic | The decider is optimizing, not satisficing |
| Goals are ill defined | The problem is computationally complex |
Klein is no escape from analysis paralysis. “Neither analysis nor intuition alone is sufficient for effective decision making” [1] — experts start from recognition; analysis is not dispensable.
5. Where intuition fails
The Power of Intuition names the failure conditions [1]; four describe software estimation: unstable or random domains where patterns cannot form (roulette, the stock market); no fast, accurate feedback — hiring judged on time-to-fill, not hire quality; a distorted experience base, the wrong experience repeated; fixation, or “bending the map”, expertise closing off the anomaly that should stop you; arithmetic — “you’re sunk if you don’t whip out the calculator”; and decisions with conflicting interests or needing an audit trail.
Organisations add barriers: rigid policies, distributed teams, turnover and constant change thin the experience base, while procedure, metrics-driven culture and IT constraints remove room to act.
6. Time pressure hits novices, not experts
Chess at 6 seconds a move against roughly 2¼ minutes under regulation [2]:
| Blunder rate, blitz | Blunder rate, regulation | |
|---|---|---|
| Masters | 7% | 8% |
| Class B players | 25% | 11% |
An interaction, not a main effect: time pressure barely touches masters and roughly doubles the error rate of weaker players. The lesson is not “people decide fine under pressure” but expertise buys immunity to it: squeezing the clock on an inexperienced team is a different act from squeezing an experienced one. The primary study is Calderwood, Klein and Crandall [3], not the trade books.
The popular version, and the half that misfires, is Thin slicing.
How solid is this?
- A strategy count, not a quality count. The 46–96% range shows experts do not compare options, not that they are right. The studies are Klein’s own, coded by his team, and the independent replications he cites are the ones that agreed.
- Descriptive, not prescriptive. RPD does not license “train people to satisfice”, and Klein says it does not transfer without consistent, fast feedback — he names stock selection, public policy and clinical psychology. Software estimation has weak feedback loops, so the transfer is a question.
- Which Klein book. Sources of Power (1998/2017) for anything with a number in it, The Power of Intuition (2004) for anything a manager should do — chapter 4 of the 1998 book is itself titled “The Power of Intuition”, which is where the confusion starts.
Acknowledgments
This content is heavily inspired by and adapted from lectures by Eduardo Miranda and David Root [4] on software project management. The structure, examples, and pedagogical approach reflect their teaching materials and frameworks.
References
- G. A. Klein,The Power of Intuition: How to Use Your Gut Feelings to Make Better Decisions at Work. New York: Currency/Doubleday, 2004.
- G. A. Klein,Sources of Power: How People Make Decisions, 20th Anniversary. Cambridge, MA: The MIT Press, 2017.
- R. Calderwood, G. A. Klein, and B. W. Crandall,“Time Pressure, Skill, and Move Quality in Chess,” The American Journal of Psychology, vol. 101, no. 4, pp. 481–493, 1988, doi: 10.2307/1423226.
- D. Root,Managing Software Development. Carnegie Mellon University / Innopolis University lecture series, 2014.
Disclaimer: AI is used for text polishing and explaining. Authors have verified all facts and claims. In case of an error, feel free to file an issue.