Radical Uncertainty

John Kay and Mervyn King

6 ideas

  1. Resolvable versus radical uncertainty

    Resolvable uncertainty covers puzzles, which have a definite answer that can be found or that fits a known probability distribution, like the odds of a dice roll. Radical uncertainty covers mysteries, where the full set of possible outcomes cannot be listed and no meaningful probabilities exist, like whether a new technology will transform an industry. Treating a mystery as a puzzle produces false precision and poor decisions.

  2. Obama's fifty-fifty call on Abbottabad

    Before the raid on Osama bin Laden's compound, President Obama's advisers gave him probability estimates ranging from about 30 to 95 percent that bin Laden was there. Obama concluded it was 'fifty-fifty,' meaning he did not know rather than making a precise estimate. The authors use the episode to show that assigning numbers to a one-off event hides disagreement and ignorance instead of resolving them.

  3. Ask 'what is going on here'

    Under radical uncertainty, decision-makers should start by building the best explanatory narrative of the situation instead of calculating expected values. They then test that story for coherence and credibility and refine it through argument and challenge from others. Good judgment is abductive reasoning toward the best available explanation, not optimization over imagined probabilities.

  4. Small-world models misapplied to large worlds

    Savage's Bayesian decision theory was meant for 'small worlds,' where every state and outcome can be specified. Economists and financiers applied it to 'large worlds,' where they cannot. Goldman Sachs's CFO said the firm saw '25-standard-deviation moves several days in a row' in 2007, which shows the models' distributions were wrong rather than that the market was extraordinarily unlucky.

  5. Ellsberg's paradox as sensible behaviour

    The authors argue this is not a bias to correct. Declining to act as if you know what you do not know is a rational response to ambiguity, so the fault lies with the theory, not the people.

  6. Reference narratives with robust, resilient strategies

    Instead of optimizing for a forecast, an organization should set out a reference narrative of how it expects things to unfold. It then chooses strategies that are robust, meaning they perform acceptably across many plausible alternative narratives, and resilient, meaning they survive shocks nobody anticipated. The aim is to avoid ruin and keep options open, not to maximize a calculated expected return.

Save and mark ideas in the app