In mining, the terms uncertainty, risk, and confidence are often used as if they mean roughly the same thing. They do not. And the consequences of blurring them are more serious than they first appear. This is not just a semantic problem. It is a decision problem. When these three layers collapse into one another, important weaknesses in understanding are easily concealed behind language that sounds reassuring but says less than it appears to.

 

The distinction needs to be kept explicit. Uncertainty is a state of incomplete knowledge. It includes limitations in geological, hydrogeological, and structural understanding, as well as the inherent variability of the ground itself. In geotechnical engineering, uncertainty is not removed simply because a model has been built. It has to be recognized, bounded where possible, and managed consciously where it cannot be reduced further.

 

Risk is something different. It is uncertainty considered in relation to consequences and objectives. It is the point at which ground behaviour becomes a decision issue. In that sense, risk is the bridge between physics and action: not just what may happen in the ground, but what it would mean if it did.

 

Confidence is different again. In this context, it is a professional judgement about whether the current level of understanding is adequate for a particular decision. Confidence is not certainty. It is not guaranteed by model sophistication, and it does not automatically increase with the volume of available data. It is a judgement about adequacy, not a claim that uncertainty has disappeared.

 

In practice, these layers are often blurred. Uncertainty is absorbed into increasingly complex models and becomes difficult to see. Risk is reduced to generic matrices that create an appearance of order without necessarily improving the quality of the decision. Confidence is then overstated under operational pressure, often because the system looks more analytical than it really is. The result is a familiar but dangerous illusion: the appearance of understanding without a proportionate increase in actual decision robustness.

 

That is why the literature has been pushing in a different direction. Modelling is only one part of decision-making, and in many cases the real limitation is not the absence of analytical capability, but the absence of explicit treatment of uncertainty. Risk matrices have similar limits. They can be useful as communication devices, but they often have weak resolution, ambiguous inputs, and a tendency to rank risks in misleading ways when used uncritically.

 

A related mistake is to assume that more data automatically produces more confidence. In reality, additional data often reveals more structure, more variability, and more competing interpretations before it improves the decision basis. That is not a failure of investigation; it is how real understanding develops. This is exactly why the idea of Value of Information matters. The question is not how much data we can collect, but whether the new information reduces the uncertainties that actually govern the decision. In slope design, this distinction has become especially important, because not all data changes reliability, and not all information improves the design choice.

 

A slope, for example, is never simply “stable with high confidence.” That phrase compresses too much into too little. In reality, the slope may remain uncertain in structural continuity, exposed to identifiable failure scenarios, and judged adequate—or not—for a particular decision at a particular stage. That distinction matters. If we do not separate incomplete knowledge from consequence, we cannot judge whether the current decision basis is truly sufficient. Reliability methods make the same point in more formal terms: factors of safety and probabilities of failure are useful only when interpreted in the context of model assumptions, data quality, and decision purpose.

 

This is why the objective of geotechnical engineering is not absolute certainty. That is neither physically realistic nor economically possible. The objective is adequacy for decision under uncertainty. That requires a more disciplined structure: uncertainty → risk → decision → monitoring → update. The logic is simple, but it is demanding. It requires uncertainty to be visible, risk to be defined in relation to consequences, decisions to be explicit, and monitoring to function not as passive surveillance but as part of a learning loop.

 

That way of thinking is increasingly aligned with current digital-twin logic as well. The real shift is not from analogue to digital, or from human judgement to automation. It is from static, report-based understanding toward closed-loop systems in which observation, modelling, interpretation, decision, and updating are connected continuously. In that environment, the distinction between uncertainty, risk, and confidence becomes even more important, not less.

 

This is also the level at which VSKY.GEO is intended to operate: not by adding abstraction, but by helping structure uncertainty, risk, and confidence so that decisions remain defensible under incomplete knowledge. In that sense, confidence should mean adequacy of understanding for the decision being made, not comfort created by models, language, or habit.

 

When uncertainty is misunderstood, confidence becomes dangerous and decisions become unnecessarily expensive. That is not only a technical issue. It is a judgement issue. And that is exactly why the explicit separation of uncertainty, risk, and confidence matters so much in high-consequence mining decisions.

​