Perspectives & Strategy
Insights
A series of technical perspectives on geotechnical risk, uncertainty, and decision quality in mining. These essays are written for technical leaders and corporate teams managing critical projects where expert senior judgment is essential for robust outcomes.
In this series
Geotechnical Judgement —
The Discipline Behind the Discipline
Geotechnical engineering is frequently defined by its technical components: soil mechanics, rock mechanics, site investigation, and complex numerical calculations. These elements form the visible structure of the field, yet they do not fully capture its essence. At its core, geotechnical engineering is not merely a study of the ground; it is a discipline of decisions made under the shadow of incomplete knowledge. The mechanism that enables those decisions is judgement.
The foundations of this philosophy trace back to the origins of the discipline itself. Karl von Terzaghi established the quantitative framework that made the field rigorous, yet he never practiced in a fully observable world. He understood that while theory was necessary, it was never sufficient—engineering required interpretation. This relationship was later formalized by John Burland through the "Burland Triangle," which illustrates that the three pillars of our craft—ground conditions, measured behavior, and applied mechanics—are only held together by the central force of experience and judgement.
This tension became more explicit through Ralph B. Peck and his "Observational Method." By updating interpretations in real-time, Peck proved that engineering is a "learn-as-you-go" exercise in managing uncertainty. In the decades since, Shunsuke Sakurai expanded this into a rigorous computational loop. Through "Back Analysis," Sakurai demonstrated that we do not just observe the ground; we use its measured response to reverse-engineer and calibrate our models, effectively turning construction into a continuous process of design hypothesis testing and refinement.
In rock engineering, Evert Hoek demonstrated that even our most data-driven parameters are constructed. Because rock mass properties cannot be measured directly at scale, they must be inferred through mapping and characterization. Across the work of these pioneers, a consistent pattern emerges: geotechnical parameters are not simply "found"—they are interpreted.
Modern practice, however, tends to obscure this reality. With the rise of sophisticated computational tools, the field has become increasingly model-driven. Numerical simulations provide a veneer of precision, but as widely recognized in modern risk theory, precision is not certainty. Today, models often reorganize uncertainty rather than eliminate it. When judgement is quietly absorbed into the analytical process, we risk a dangerous shift: from questioning what we do not know to simply trusting what the model says.
This shift creates a "blind spot" in the traditional Factor of Safety (FS). A calculated FS of 1.5 based on high-quality data represents a different reality than a 1.5 based on sparse, "guessed" parameters. As Allen Marr has noted, even the most experienced teams are susceptible to cognitive biases—anchoring to initial data or becoming overconfident in familiar simulations. When judgement becomes implicit and invisible, it goes unchallenged. Failures are rarely the result of a lack of analysis; they are more often the product of unchallenged assumptions and the normalization of uncertainty.
As the discipline evolves, the role of judgement is becoming more critical, not less. Increasing data does not eliminate uncertainty; it introduces more interpretations. In this environment, the "Geotechnical Digital Twin" reframes the system—not merely as a static model, but as a continuous decision process that enforces the loop of observation, interpretation, and update.
This is the role VSKY.GEO is structured around: independent geotechnical judgement applied to high-consequence decisions, where uncertainty cannot be eliminated but must be made explicit, challenged, and continuously reinterpreted as knowledge evolves.
The pioneers of the discipline understood that while the ground is governed by physics, it is managed through judgement. In modern, model-driven workflows, this function can become obscured. Re-establishing it - explicitly, independently, and continuously, which is essential to maintaining a valid basis for decision as understanding evolves.
The Geotechnical Digital Twin —
A Decision System, Not a Model
In the modern mining landscape, the term “Digital Twin” is ubiquitous, yet it remains one of the most poorly defined concepts in the industry. It is often used to describe a high-resolution 3D visualization or a real-time data dashboard. However, in high-consequence geotechnical environments, these are merely digital artifacts. They represent the state of the ground, but they do not manage the risk of the ground. Most slope failures are not caused by a lack of modeling; they are caused by the widening gap between the model and the decision.
To understand the true value of a Geotechnical Digital Twin, we must categorize digital representations into three distinct levels of maturity. First is the Digital Model, a static snapshot of the system. Second is the Digital Shadow, which updates continuously via sensor data but remains a passive reflection. The third, and most critical, is the Digital Twin. What separates a Twin from a Shadow is the integration of state, prediction, and explicit decision logic. This is achieved through a Digital Thread, a continuous and traceable flow of information that connects initial geological investigation to real-time operational response, ensuring that no piece of evidence is left isolated in a silo.
A true Geotechnical Digital Twin is not a replica of a slope; it is a continuously updated, uncertainty-aware decision system. It operationalizes the loop that has defined the discipline since its inception: Observe → Interpret → Update → Decide. In modern practice, this update is increasingly powered by Bayesian Data Assimilation. In this framework, the twin acts as a continuously updated statistical engine; every new sensor reading doesn't just provide a point in time—it mathematically recalibrates our initial assumptions, reducing the envelope of uncertainty and refining our predictive accuracy as the project evolves.
This reframing shifts the focus of the geotechnical department. Periodic, manual model updates are replaced by continuous ownership of the geotechnical state. Most importantly, this allows for more targeted and adaptive design decisions. By leveraging real-time confidence in the ground’s behavior, engineers can move away from excessive, “worst-case” conservatism. This improves both economic efficiency and the targeting of stabilization measures, reducing the tendency toward blanket conservatism.
The system must be built to answer one fundamental question relentlessly: What has changed—and does it require action?
Most current systems fail this test. They provide a window into the slope’s behavior but offer no bridge to the engineer’s response. They record the movement, but they do not guide the actions. A Geotechnical Digital Twin is therefore not defined by its precision or its graphical complexity. It is defined by its utility. Its success is measured by whether it changes a decision as reality evolves—turning the ground from an unpredictable challenge into a system that can be interpreted and acted upon with clarity.
The geotechnical digital twin remains an evolving concept in rock engineering. This is the level at which VSKY.GEO operates: linking observation, interpretation, and modelling directly to decision-making in high-consequence geotechnical systems.
Uncertainty, Risk, and Confidence:
Stop Mixing Them
In mining, the terms "uncertainty," "risk," and "confidence" are frequently used interchangeably. This is not merely a semantic oversight; it is a fundamental decision error. When we fail to distinguish between these three layers, we create a structural ambiguity, where critical vulnerabilities are masked by optimistic language.
To manage high-stakes environments, we must enforce a rigorous separation:
• Uncertainty is a state of incomplete knowledge. It is the measure of the "known unknowns"—the gaps in our geological, hydrogeological, or structural data. It represents the inherent variability of the ground that no amount of modeling can fully eliminate.
• Risk is uncertainty evaluated against consequences. It is the bridge between physics and decision-making. Risk asks: If our uncertainty is high, what is the potential cost to safety, equipment, or the mine's economic value?
• Confidence is a human and professional judgement. It is the answer to the definitive question: Is our current level of knowledge sufficient to justify this specific decision?
In current practice, these layers are often blurred. Uncertainty is hidden within the "black box" of complex models, risk is oversimplified into generic 5x5 matrices, and confidence is frequently overstated to satisfy operational pressure. This creates an illusion of understanding that leaves decisions fragile.
A common fallacy is that more data automatically leads to more confidence. In reality, data without interpretation often increases perceived uncertainty by revealing more complexity. True confidence is not a function of data volume; it is a function of Value of Information (VoI)—knowing which uncertainties actually matter to the risk profile and which do not.
A slope is never simply "stable with high confidence." In reality, it is uncertain in structure, exposed to specific risks, and judged adequate for a decision. The distinction matters because if we do not separate the lack of data (uncertainty) from the potential for failure (risk), we cannot accurately determine if we are ready to proceed.
The objective of geotechnical engineering is not to achieve absolute certainty—that is a physical and economic impossibility. The objective is adequacy for a decision under uncertainty. This requires a consistent structure: Uncertainty → Risk → Decision → Monitoring → Update.
This is the level at which VSKY.GEO operates: structuring uncertainty, risk, and confidence so that decisions remain defensible under incomplete knowledge, and ensuring that “confidence” reflects adequacy of understanding rather than assumption.
When uncertainty is misunderstood, confidence becomes dangerous, and decisions become unnecessarily expensive.
Do Not Delegate Judgement to Models
Even well-constructed geotechnical models are often trusted beyond the limits of the knowledge they represent. In modern practice, we have partially replaced the messy, unpredictable reality of the field for the clean, colorful outputs of numerical simulations. We produce Factors of Safety to three decimal places and stress-strain curves that suggest total control. However, a model is not reality; it is a simplified hypothesis about reality, constructed from limited data and bound by the limitations of mathematical assumptions.
As Starfield and Cundall (1988) famously argued, geotechnical engineering is a data-limited discipline. Unlike structural or mechanical engineering, where materials are manufactured to a precise specification, we work with materials provided by nature. heterogeneous, hidden, and often hostile to simplification. They observed that the true art of modeling lies not in complexity, but in determining which aspects of geology are essential and which can be safely ignored. If we build models that are as complex as the real world, we have not solved the problem; we have merely duplicated the confusion in a digital format.
The most dangerous phase of this process is the calibration illusion. In the industry, we often confuse calibration with validation. Calibration is the process of tuning the math to match historical field data, essentially adjusting parameters until the model fits what has already happened. This is what mathematicians call the Inverse Problem: because the ground is a partially observable system, conflicting geological interpretations can produce identical calibrated outputs. A model that matches yesterday’s displacements is not necessarily true; it is simply one of many possible mathematical fits.
In rock engineering, this problem is compounded by the DIANE nature of the ground: it is Discontinuous, Inhomogeneous, Anisotropic, and Not-Elastic. Numerical models do not inherently capture scaling effects or latent relationships, such as pore pressure buildup in specific joints or the degradation of rock bridges, unless we explicitly assign them. As Herbert Einstein noted, the double uncertainty of the ground means it acts as both the structure and the load. If the engineer does not deeply understand these physical drivers, the model remains a conditional representation, where uncertainty is not managed, but merely hidden behind a veneer of computational precision.
The real question we must ask is not:
Is the model showing a stable slope?
It is: Is our decision robust to what the model cannot see?
Models should never be treated as providers of absolute answers. Their role is to act as an experimental test bench, to explore sensitivities, test what-if scenarios, and expose the drivers of failure. A strong modeling process expands our understanding of risk; it does not compress it into a single, comfortable number.
This is the level at which VSKY.GEO operates: ensuring that models are used to test assumptions, expose uncertainty, and support defensible decisions under incomplete knowledge, rather than to justify predetermined conclusions.
Beyond Conservatism: Finding the Limit
in Mine Design
In the global mining industry, we are obsessed with production efficiency. We optimize truck cycles, automate drills, and refine processing recovery to the second decimal point. Yet, most mining value is not created during production. It is created—or quietly lost—much earlier, in the gap between the geological model and the final design.
Operations are forced to optimize within a system already structurally defined by the geotechnical design: slope angles, stope geometries, and extraction sequences. As Ted Brown famously emphasized, these are not merely technical parameters; they are the primary drivers of a mine's economic engine. Because the geometry of a mine is the foundation of its Net Present Value (NPV), a change of just a few degrees in a pit slope or a few meters in a stope span can represent a swing of tens of millions of dollars.
Despite this, geotechnics is rarely viewed as a value driver. Instead, it is often treated as a compliance constraint or a safety cost center. This is a profound strategic error. As Dick Stacey argued in his work on design efficiency, an engineer’s duty is not just to prevent failure, but to achieve the optimal use of resources. Over-design is not "safe engineering"—it is a failure of resource stewardship that locks up millions of dollars of mineable metal in oversized pillars and unnecessary stripping.
Value loss in mining rarely appears as a headline-grabbing failure. Instead, it manifests as chronic, systemic conservatism. When uncertainty is not properly interpreted, the default response is to "over-design." This creates a massive, hidden "insurance premium" paid by the project in the form of:
• Flatter slopes that increase waste stripping and defer ore access.
• Oversized pillars that lock up millions of dollars of mineable metal.
• Excessive support that slows down development rates and inflates CAPEX.
Brown argued that "Optimal" is not synonymous with "Risky"—rather, it is the precise application of engineering. This aligns with the logic of the Global Industry Standard on Tailings Management (GISTM), which requires an explicit, structured understanding of risk. The goal is to reach the ALARP (As Low As Reasonably Practicable) risk level without drifting into wasteful over-engineering that destroys the project's viability. This requires making uncertainty explicit rather than embedding it implicitly in conservative assumptions.
Modern technology improves our visibility, but as Baecher and Christian point out in their work on reliability, data alone does not create value. Value is only unlocked through the Value of Information (VoI)—the point where new data actually changes a design decision. Additional data creates value only when it either changes the decision or strengthens confidence in its adequacy.
This is the level at which VSKY.GEO operates: positioning geotechnics as a driver of value under uncertainty, and ensuring that design decisions reflect explicit interpretation of risk rather than implicit conservatism.
Value is not only destroyed by failure; it is quietly lost in design decisions made too early, before the ground is understood. Maintaining alignment between evolving understanding and design decisions is essential to preventing this loss and capturing available value.
Why Remote Advisory Works —
When It Shouldn’t
At first glance, remote geotechnical advisory seems counterintuitive. Mining is a physical industry; the ground is observed, felt, and heard in the field. However, a critical distinction must be made: while data is collected at the rock face, high-consequence decisions are made in the space of interpretation. As the industry moves toward digital integration, we are discovering that distance is not a barrier to safety—it is often a prerequisite for clarity.
The most significant turning points in a project—model reviews, risk acceptance, and the challenging of design assumptions—are cognitive acts, not physical ones. In fact, being physically present on a site can sometimes be a liability to objective judgement. As Nobel Laureate Daniel Kahneman explored in his research on Noise, experts are highly susceptible to "environmental bias." A site-based team is under immense, constant pressure to meet production targets. This proximity can lead to a gradual normalization of deviance, where subtle warnings in the ground are overlooked because they have become part of the daily background noise.
A remote advisor introduces cognitive independence. By operating outside the immediate operational and social pressures of the mine site, an advisor provides what Kahneman calls "decision hygiene." This distance allows for:
• Challenging "Local Logic": Questioning the "way we’ve always done it here" and ensuring risk ranking follows transparent, logical structures rather than habit.
• The "Cold Eye" Review: Looking at "Worst-Case" scenarios that a production-focused team may be psychologically inclined to downplay.
• Unbiased Scrutiny: Ensuring that confidence levels are based on interpreted evidence—such as real-time sensor readings and production metrics—rather than the need to meet a monthly quota.
Critics often argue that remote advisors lack the direct observational familiarity of the ground. However, as John Burland demonstrated with the stabilization of the Leaning Tower of Pisa, the key to success is not just more data, but a robust mental model of the ground's behavior. In an era of high-frequency InSAR, automated logging, and real-time piezometers, the "observer" no longer needs to stand in the pit to see the reality. They need the uninterrupted time to interpret it—a luxury rarely afforded to site-based personnel.
Remote advisory is not a replacement for site-based engineering; it is an independent layer of interpretation within the decision process. It aligns naturally with the loop: Observe → Interpret → Update → Decide. This is the level at which VSKY.GEO operates: providing independent, structured interpretation that strengthens decision clarity under operational pressure and evolving uncertainty.
Location is secondary. The most critical decisions are not made where the rock is exposed; they are made where uncertainty is most clearly interpreted.
Independent Geotechnical Advisory for Strategic Mining Decisions.
Strengthen your technical judgment with independent review and senior expertise. Contact VSKY.GEO to discuss supporting your next high-impact decision.
Contact: vsky.geo@outlook.com