Measured, Indicated, Inferred (and alleged)

I’ll clear up the question first. We don’t grow roses in our garden so you’ll need to accept a grevillea flower instead!

What’s in a name??That which we call a rose/ By any other name would smell so sweet.

I hate to disagree with a literary masterpiece but names, and words in general, do matter. I think even that old childhood mantra ‘sticks and stones’ is probably a bit discredited by now. These days we pay serious attention to matters of speech craft and their impact. In some cases the import allowed to words is extreme. And that can be a problem.?

Words carry meaning and sentiment. The way we individually perceive their meaning is a fascinating field in the cognitive sciences. Language, it turns out, plays a pretty important role in shaping our thinking and our perceptions of the world. Linguists can struggle to convey meaning where languages do not fully overlap. Even when the translation is a close conceptual approximation it is unlikely to have the same impact. Language, culture, sentiment all tied together in a tangled web.

This is starting to be Gedankenkarussell…

Why has this grabbed my attention? It comes back to risk, risk perception, risk appetite. It comes back to how we create a shared understanding for all stakeholders. I keep circling around the drain hole, pulled deeper and deeper by the suction of our jargon.?

Jargon. Jargon is great! Where would we be without jargon? Without our own professional sub-language whose meaning is only known to initiates??Why, the ill informed and illiterate masses might start seeing past those fancy words and recognise that too many of us hide behind complex words when simple words would suffice (I suspect I’m guilty here).?Equally, those non-specialists might start to recognise that jargon is like a fortified wall defending some professions. If you must learn a new language just to participate in the conversation, jargon can be a formidable barrier to entry.?

So when I tell you something occurs as poikioblastic porphyroblasts I’m either assuming a pretty high level of share knowledge or I’m trying to confound you while looking erudite.?

Yes, jargon has its uses. Amongst people with similar or related training and experience it can be a great shorthand and bypass a load of less effective language. But, what happens when jargon collides with common language? You know, when a jargon word with one definition is the same as a common language term??As a resource professional this is an almost daily occurrence. Worse, some of those common language terms that have been absorbed into our technical jargon have taken on differing connotations for different professionals.?

Like ‘measured’, ‘indicated’, and ‘inferred’ (and the unmentioned ‘alleged’). Like ‘proved’ and ‘probable’

As a non-specialist if someone told me they had a ‘proved reserve’, I would fully expect that to mean it was a foregone conclusion beyond a shadow of doubt - black and white! Those tonnes, that grade, the metal was ‘proved’ to exist. Thus, if I invested in your ‘proved’ reserve I would fully expect to see that metal delivered to the market. The non-specialist connotation of ‘proved’ is quite specific. Particularly so in those of a less scientific and engineering mind. Look at how that word ‘proved’ is so often used. “The science is proved”. Once some science concept takes hold in the common perception that paradigm becomes accepted fact, unalterable and unchallenged (and woe betide anyone who challenges!)

That term ‘proved’ has a slightly different meaning to a geologist and engineer. We are a bit wary of saying something is a foregone conclusion.?In the deepest recesses of our minds we -know- that sometimes bad (and good for that matter) things happen. We have a subconscious perception of how different things could be. I suspect, and it’s something I’d love to test someday, that many geologists and engineers, when asked what they mean by a ‘proved reserve’ would encompass those numbers with error bars - and I suspect they would say something along the lines of +/-10%.?Interesting. 10% precision or error is a great example of a shared fantasy.?Not only in mining but in many other fields that deal with uncertainty.?

For example, in their book ‘Noise’, Kahneman, Sibony and Sunstein quote a study they conducted of insurance underwriters and claims adjusters.?These learned gents first interviewed over 800 CEOs and executives asking what level of variation they would expect in the judgement of experts. The most common answer? +/-10%, followed by those with a bit more tolerance at +/-15%. Those are fairly wide bands by some measure but a 10-15% variation would usually be manageable, if with some anxiety.?

With this foreknowledge in hand, our intrepid researches did some… research. For underwriters the -median- range for policy quotes was 55% on the same policy by different underwriters. The adjusters fair slightly better at 43%. And those were the -median-… the variance for more half of the policies was -greater- than 55%….

When I first read those numbers I was struck by the likely similarity to our reporting of mineral resources and ore reserves.?We have a pool of experts.?Those experts are assessing risk under conditions of uncertainty. They have different experience and expertise. They are dealing with different commodities, deposits, markets. We have executives and investor stakeholders who have their own beliefs around the fallibility or otherwise of those experts.?And +/-10% is a recurring performance limit I see over and over and over again in reconciliation reporting and analysis. If you reconcile within 10% people are much less likely to come searching for ‘answers’ than if you reconcile outside that band.

But… like lognormal distributions, inverse-distant-squared and top-cutting at 97.5%, there is nothing in the least bit special or magical about +/-10%. It is a shared fantasy with no basis in any sort of analysis.?

Strange then that the magic number (10%) has crept it’s way into our perceptions around resource classification. Ha! I even use it myself when I try to communicate outside the hallowed halls (ivory towers?) of resource estimators.?Over time and on the recommendations of several well regarded experts proved and probably, measured and indicated, have become conflated with +/-10% and +/-15% - the sophisticated amongst us also apply a volumetric get-out-of-jail-free-card by saying +/-10% over a quarter and +/-15% over an annual production volume.

Those bands are ok after a fashion I suppose. We can treat them as ‘specification limits’ if nothing else. Even then however, they raise some interesting questions.

Firstly is 10% what an investor thinks when they hear something is ‘proved’? Have we calibrated our ideas with those of our stakeholders? What about the more nebulous ‘probable’? What does imply? The common use of the word is usually a reflection of a likelihood, as in ‘more probable than not’ meaning that something is more likely to happen than not happen. Some would read that as 51% vs 49%. Others 60-40 and so in.

Secondly, how do we know? How do we measure this noise, this precision and error? The insurance example was relatively simply.?Present the same request for a quote to multiple underwriters, ensure they have the same information and see what happens. Should we do the same? Should our resource estimates take a Delphic approach? Should we enlist the wisdom of crowds and check in the range and central cases? In many ways that appeals to me, albeit that the practical difficulties are monumental.

We -could- simulate but the tools are not entirely complete. I have hope though. Mind you, a simulation can also fail to capture the full uncertainty - particularly from a structural geology perspective. We have a bias to focus on grade when discussing variation - we must. It forget the geometric and volumetric aspects!

By the way.. don’t tell me reconciliation is the answer to understanding the precision of an estimate. At best, reconciliation tells us how well we predicted the tonnes and grade of the material we mined and sent to the ore treatment plant.?It tells us nothing of the ore sent to waste. It tells us nothing of the ‘ore’ left in the walls as under-break. It tells us nothing of the ‘big picture’ resource and the inter-relationship between the mineralisation and the pit optimisation. And using the ‘reconciliation’ to ‘improve’ the resource is a bit like using past share price performance as a predictor of future performance. Yeah, it might work, but if the nature of the geology changes (and doesn’t it always?) you are just as likely to add to the error as you are to derive better predictions… so just be a little careful about using those machine learning approaches folks. I’d hate to see you bias your outcomes because the ML was fed unknowingly biased data (hmmm that’s never happened right?)

Back to those five (or 6) words.

Thirdly, and this is a concept I think our financiers will struggle to grasp, is the confidence interval for measured/proved?on a narrow-vein, high nugget gold deposit as it is on a coal deposit or iron ore? What about lithium? Phosphate? If we say ‘proved’ is +/-10% and we figure out a way to quantify that variation, will there ever be a ‘proved’ reserve for a commodity where the metal price has extreme volatility? To flip this around, is ‘proved’ the same in gold and in a porphyry copper? If so, why? If so, does that imply the investment risk is the same?

Words fail me… I mean that literally. I reckon that continuing to use words with both a common and jargon definition we are just asking for more trouble. Let’s face it, even those words (in their jargon) ‘proved’ and ‘probable’ are getting a bit dated now, stretching back more than 50 years, like your’s truly they are no doubt getting a bit world weary (or is that weltschmerz?). ‘Measured’ and ‘indicated’ are in their late 30’s and the kids today speak a different language!?

So let’s move on and modernise. If we are going to be bold, let’s be downright disruptive.?If things are going to change to catch a few competence outliers, let’s make the change meaningful and give it a solid chance of working.

What am I proposing? Nothing specific at this stage. It needs more thought and more debate. More widespread fertilisation. I can’t help but think that as we see rapid increases in the uptake of automated and instantaneous data capture, automation in data analytics, automation in estimation and simulation we need to consider the consequences when it comes to assessing resource and reserve risk.?

I’m particularly keen on seeing that risk assessment address the influence of system constraints and the complex interactions of interdependent systems. My intuition tells me there’s a lot of value to be lost or gained by understanding that stuff. Look, for example at what happens when you exceed r=3.56995 in a logistics map. Our mining operations are, after all, non-linear systems!

Let’s look at reporting confidence intervals and alternative plans and outcomes under different external scenarios. Let’s look at how the human element might affect performance. Let’s design an approach that highlights what -must- go right and what happens when it goes wrong.

Too far? Well possibly. But bold steps require bold vision, not simply asking someone to log 50 hours of professional development every year and pass an assessment exam. That’s a bit trivial and doesn’t address the root cause of the problem. Besides, who gets to set the rules around a PD hour? Surely not those that profit from delivering PD…

Postscript. The ‘alleged’ classification is not part of any officially recognised reporting code. It’s a term that is sometimes used as a derogatory comment when observing estimates that are so clearly inflated or otherwise unbelievable that the reviewer can only believe the estimator is allegedly ‘competent’.Measured, Indicated, Inferred (and Alleged).?