04 May 2022
Kobold Metals co-founder and CEO Kurt House speaks on the importance of reducing uncertainty during mineral exploration
By Matthew Parizot
As mineral deposits become more rare and harder to mine, mineral exploration has become increasingly reliant on quality data collection and interpretation to find deposits.
The final keynote event of the CIMBC22 conference, featuring a presentation from Kobold Metals co-founder and CEO Kurt House, focused on what he believed is one of the dangers of this new reliance on data: misinterpretations and spurious correlations.
For those who are not well-versed in statistics, when interpreting data, any single instance of a correlation may not be enough to make an accurate statement across the board. Singular borehole data with a high level of confidence for one deposit can’t be used to accurately predict an entire mineral claim. Making these decisions with incomplete data goes against what House said is the true objective of exploration.
“[Finding economic ore bodies], that’s a motivation, but it’s not the central objective of the exploration process. [Rather], it’s to continuously reduce uncertainty by making optimal measurements,” House said.
Optimal measurements, according to House, are the ones that result in the largest increase in model predictive power per unit of exploration effort. By doing this, explorers can focus their attention on the areas that will give the most important information to the model analyzing the data and making predictions on the composition of minerals in different areas.
“Getting people to the right location to make new measurements is really expensive. This is the high marginal cost of data,” House said. So, the whole game is to make those optimal measurements. How do you get people to the right locations, to the places where the new data will decrease uncertainty the most?”
He presented an example where, given several models of different deposits in an area, it is preferable to gather information around a deposit with a higher level of uncertainty, rather than one that is more clearly defined. By doing so, you can increase the confidence of the entire analytical model and increase accuracy across the board.
“Basically, what you’ve done is taken a highly uncertain prediction and turned it into a high-confidence training point. That’s potent. What’s really cool about it is, [since] you’ve adjusted the coefficient of the whole model, you can apply it over the entire domain. You can look 100 kilometres away to a totally different area and the models change a lot.”
Improving these models to define areas of mineralization more accurately is of course a valuable effort to exploration companies, but also an important step in increasing the supply of critical minerals needed to fuel the planet’s green energy transition.
“In order to address climate change, we need to electrify the economy. Just electrifying the light duty vehicle fleet requires enormous growth in copper, nickel, cobalt and lithium,” House said. “There’s an incredible amount of work to do and climate change does not get solved without massive electrification. That requires a massive amount of materials. That’s why we’re doing what we’re doing.”