Why probability beats prediction in subsurface exploration

The ground is read from almost nothing

Consider how little is actually known. A deposit might be sampled by a handful of drillholes across hundreds of metres of rock, supported by geophysics that infers structure rather than seeing it. Most of the volume is never touched. Everything between the holes is estimated.

From that sparse, uneven evidence, someone has to commit millions to a drill programme. A model that answers with a single point prediction pretends the gaps are filled. They are not. The honest output is a distribution: a range of outcomes, each with a likelihood attached.

Prediction hides risk. Probability prices it.

A confident prediction feels reassuring and behaves badly. It gives you nowhere to put the uncertainty, so the uncertainty shows up later, in a dry hole or a downgraded resource.

A probability does the opposite. It carries the risk on its face. It says this target is more likely than that one, this estimate is well supported and that one is thin, this is where more data would change the picture most. Every one of those statements helps you decide where to spend. Prediction asks for faith. Probability supports judgement.

Judgement is the point

Exploration is capital allocation under uncertainty. The job is not to be told an answer. It is to weigh the odds and act.

That is why probability wins. It does not replace the geologist, the investor or the decision. It arms them, with the one thing a point prediction leaves out: an honest measure of how likely it is to be right. In ground you cannot see, that measure is everything.