Geotechnical Judgement —
The Discipline Behind the Discipline

Geotechnical engineering is the discipline that brings together geology, soil and rock mechanics, investigation, monitoring, and modelling to understand how the ground is likely to behave and to support engineering decisions under uncertainty. Its technical methods are fundamental, but they are not the discipline itself. They are the tools by which incomplete and evolving knowledge of the ground is interpreted, challenged, and translated into decisions about design, excavation, stability, and risk.

 

That distinction matters. The ground is never fully revealed in advance. It is encountered in fragments: through boreholes, mapping, testing, monitoring, exposure during excavation, and sometimes through behaviour that was not anticipated. In that sense, geotechnical engineering has never been only about calculating response from known inputs. It has always been about making consequential decisions before the system is fully understood.

 

That is why judgement sits so close to the center of the discipline. Not as a substitute for analysis, and not as intuition detached from evidence, but as the mechanism that connects incomplete knowledge to action. The role of the geotechnical engineer is not simply to process data. It is to decide what the data means, what remains uncertain, and whether the basis for design or operation is strong enough for the decision at hand.

 

The founders of the discipline understood this clearly. Terzaghi gave geotechnical engineering its scientific backbone, but he did not practice in a world of complete observability. He knew that theory was essential, but he also knew that natural ground could not be reduced to theory alone. Engineering required interpretation. Later, Burland captured this elegantly in the geotechnical triangle: ground profile, behaviour, and modelling at the vertices, with empiricism, precedent, experience, and judgement at the center. It remains one of the most honest descriptions of how geotechnical engineering actually works.

 

Peck pushed this further through the observational method. His essential argument was that in geotechnical engineering the first interpretation is rarely the last one. We proceed with a working understanding, observe what the ground actually does, compare that behaviour with expectation, and revise if necessary. That is more than a practical method. It is a statement about the nature of the discipline itself. We do not eliminate uncertainty before acting. We act with discipline inside uncertainty, and we improve the decision basis as reality becomes clearer.

 

The same logic runs through rock engineering. Sakurai and others showed how back-analysis can turn excavation into a continuing test of interpretation. Hoek, in a different but equally important way, showed that rock-mass properties are not simply measured and retrieved. At engineering scale they are inferred through mapping, classification, characterization, and experience. Before the model begins, judgement is already present in the choice of domains, structures, parameters, and mechanisms. The calculation may look rigorous, but much of what enters it has already been interpreted.

 

Brown adds another important layer to this tradition. In his later work, he treated risk assessment and technical review not as peripheral activities, but as formal parts of rock engineering practice. That matters because it shifts judgement from something merely personal or tacit into something that can be structured, challenged, and updated. In Brown’s framing, the hard problem is not only how to calculate, but how to reason under uncertainty as evidence evolves, and how to subject important designs and assumptions to competent review before confidence hardens too early.

 

This is why I have never found it convincing when geotechnical engineering is presented mainly as a computational exercise. Modern tools are powerful, and they have unquestionably improved practice. We can monitor more, model more, visualize more, and update faster than at any previous point in the discipline. But these advances do not remove the central difficulty. They only change its form.

 

The difficulty remains the same: the ground is heterogeneous, partly observed, and never fully captured by the model. A result may look precise, but the understanding behind it may still be conditional. A factor of safety of 1.5 does not mean the same thing when it rests on strong characterization and a well-resolved mechanism as when it rests on sparse data, uncertain structure continuity, or a weak hydrogeological concept. The number may be identical. The decision basis is not.

 

That is often where the real engineering problem lies. Geotechnical failures are rarely explained by lack of analysis alone. More often they emerge from something quieter and more familiar: assumptions that were not challenged hard enough, uncertainty that became normalized, or confidence that gradually outran the quality of the underlying understanding. In practice, many failures are not failures of calculation. They are failures of interpretation, or failures to revisit interpretation when new evidence begins to point elsewhere.

 

Judgement is central here, but it is not automatically reliable. It can be distorted by anchoring, habit, overconfidence, organizational pressure, and attachment to familiar models. Experience helps, but it does not remove these risks. In some cases, experience makes them harder to detect because the engineer becomes more comfortable with the structure of the problem than with the possibility that the structure itself may be wrong. That is why judgement cannot remain hidden inside the process. It needs to be made more explicit, more testable, and more open to challenge.

 

This, to me, is one of the most important shifts now emerging in geotechnical engineering. The role of judgement is not shrinking as modelling and data improve. It is becoming more important. More data creates more possible interpretations. More sophisticated models introduce more assumptions, not fewer. More integrated digital workflows increase the risk that confidence becomes embedded in the system before it has been properly examined.

 

That is where the idea of the geotechnical digital twin becomes genuinely useful. Not as a prettier model, and not as another digital label, but as a way to structure the continuous loop that geotechnical engineers have always had to manage: observe, interpret, update, decide. In that sense, the twin does not replace judgement. It gives judgement a more explicit, auditable, and continuously updated place in the decision system.

That is also the logic behind VSKY.GEO. The point is not to compete on producing more calculations or more reports. It is to apply independent geotechnical judgement where decisions are costly, uncertainty is real, and assumptions need to be made visible before they harden into design or operating practice. In many situations, the greatest value is not in adding another layer of analysis, but in introducing more rigor into the decision-making process itself. It is in clarifying what is actually known, what is only assumed, and what that means for the next decision.

​

The pioneers of the discipline understood something that remains easy to forget: the ground is governed by physical reality, but geotechnical engineering is governed by decisions made before its complexity is fully revealed. That is why judgement is not an optional human addition to an otherwise technical process. It is one of the core disciplines within the discipline.

​