Definition
Decision under uncertainty is the act of choosing when you do not have full information, sufficient time, or a reliable model — which Rosling argues is the condition of almost every real-world policy decision and most personal ones. The honest position is to admit the uncertainty and adopt a procedure that copes with it rather than pretend it is absent.
The procedure Rosling endorses: collect more data quickly, act small while you collect, evaluate, repeat.
Why it matters
How it works
Three operative moves. First, quantify what you can — give every parameter a range rather than a point estimate. The exercise often reveals that one or two unknowns dominate the rest, so data-gathering effort can be focused. Second, identify the irreversible choices in the decision tree and avoid making them under uncertainty if any alternative exists; reversible choices are cheap mistakes. Third, set a review trigger — a date or a metric threshold — at which you re-evaluate. Decisions without review triggers ossify into the default and stop being decisions.
Rosling's strongest recommendation is to refuse the false binary "act decisively or wait until certainty." Both are usually unavailable. The correct frame is "what is the smallest action consistent with my best current understanding, and how fast can I learn whether it worked?" Many policy failures (and personal ones) collapse to a refusal to ask that question.
A useful diagnostic: when proponents of a decision dismiss the request for review triggers as 'lacking confidence' or 'undermining commitment,' assume the decision is being made in pretend-certainty mode and the next step is to reintroduce uncertainty into the conversation explicitly.