Why and How Uncertainty Should Be Measured
- David Fox
- Feb 17
- 3 min read

Debunking the Myths of Measuring Uncertainty
I often hear arguments against measuring uncertainty, which usually fall into two camps:
“Well, it’s just uncertain—no one knows what will happen.”
“There are too many variables to make an accurate assessment.”
These arguments may sound reasonable, and you might have encountered them—or even used them yourself. Fortunately, they are both incorrect. Let me explain why.
What Measuring Uncertainty Does Not Mean
Measuring uncertainty does not involve using a measuring tape, callipers, or accountants. These tools provide precise, singular values such as 1,320mm, 6.37mm, or $3,765—useful when measuring a specific, fixed attribute. However, real-world uncertainty often involves variation, such as different cable lengths, manufacturing tolerances in glass thickness, or fluctuating airline ticket prices.
The Purpose of Measuring Uncertainty
When we measure uncertainty, we aim to improve our knowledge and understanding of a situation. Simply stating that something is “uncertain” tells us nothing. Even classifying it as “medium uncertainty” is unhelpful—does “medium” mean twice as much as “low” or half as much as “high”?
Instead, uncertainty should be expressed as a range. For example, if five samples of glass thickness measurements vary between 4.1mm and 5.5mm, then we now have a useful range rather than a vague label like “medium uncertainty.” Even with a small sample size, this approach improves our understanding significantly compared to having no data at all.
The Role of Uncertainty in Science and Engineering
In science and engineering, measurements often contain a degree of tolerance—another way of expressing uncertainty. When estimating average human height, we don’t measure just one person and declare, “All people are this height.” Nor do we attempt to measure every individual. Instead, we measure a representative sample to establish a height range and an average estimate for the population.
Similarly, in risk management, measuring uncertainty is about improving understanding—not achieving absolute precision.
A Historical Example: Eratosthenes and the Earth’s Circumference
A fantastic example of measuring uncertainty comes from Eratosthenes, a Greek scholar who lived between 276 and 194 BC. By making two simple observations, he was able to estimate the Earth’s circumference to within 3% of its true value—an incredible feat for the time. If an ancient Greek could estimate the size of our planet with remarkable accuracy, we can certainly measure the uncertainty surrounding a business or project event.
How Probability and Frequency Differ
When quantifying risk, we not only assess the range of uncertainty but also the probability of occurrence. Probability, however, often confuses people.
I once had a senior engineer argue that a 10% project risk meant it was guaranteed to happen over the next two years. Can you spot the flaw in his thinking? He was confusing probability with frequency.
Frequency measures how often an event occurs within a set timeframe. It applies to repeatable events, such as a light bulb failing once every 100 days (1% failure rate).
Probability expresses confidence in a one-time event occurring. For example, the risk of Lifts Are Us Ltd. going bankrupt during a project’s duration is not a repeatable event—it is a singular risk with an estimated probability.
In health and safety, frequency-based risk is used because it is derived from historical industry data. However, in strategic, enterprise, and project risk management, we use probability to assess the likelihood of a single event occurring within a given timeframe.
Why Quantifying Uncertainty Matters in Risk Management
When quantifying risks, we assess both:
The probability of occurrence (e.g., “There is a 30% chance that Lifts Are Us Ltd. will go bankrupt during the project.”)
The range of potential impact (e.g., “If this occurs, the financial impact will be between $100,000 and $750,000.”)
Compare this to a qualitative risk report that merely states, “This is a medium risk.” Which provides more actionable information?
Turning Data into Actionable Insights
Quantified risk analysis enables organisations to make informed decisions. For example:
If a project board deems a 30% probability of financial loss between $100,000 and $750,000 unacceptable, they can develop mitigation strategies to reduce either the likelihood or the financial impact.
Decision-makers can now weigh the cost of mitigation against the potential financial exposure, leading to more strategic and cost-effective risk management.
The Next Step: Monte Carlo Analysis
Understanding individual risks is just the beginning. In my next article, I’ll explain Monte Carlo analysis—a powerful modelling technique that aggregates individual quantified risks to give a clear picture of total business or project risk exposure.
Stay tuned! Follow or subscribe to ensure you don’t miss the next article in this series.