Uncertainty, in testing, describes things that can't be measured or quantified, unlike risks, which can be assigned some kind of probability. The distinction comes from the economist Frank Knight, and is sometimes called Knightian uncertainty.
Why it matters: People tend to prefer risks they can evaluate, so risk-based testing can drift towards quantifiable risks and away from open questions. Treating testing as the work of exposing uncertainty helps cover the blind spots a risk model misses.
In practice: Testers look for what isn't yet known, through exploration, questioning and exposing assumptions, then turn uncertainty into understood risk where they can. Good test reporting covers assessed risks, the assumptions behind them, confidence levels and the areas still uncertain.
Examples:
- Reporting not only assessed risks but also the areas where the team is still uncertain.
- Spending exploratory sessions on a new integration where no data exists to estimate risk.
Common pitfalls:
- Reporting a risk assessment as if it were certain.
- Testing only what can be measured, and ignoring open-ended questions.
What's the difference between risk and uncertainty? A risk can be measured or given a probability. Uncertainty can't. Testing often works to turn uncertainty into risk that teams can reason about.
