Can crowdsourcing or algorithms find new deposits? It’s certainly a challenging question for exploration geologists particularly those that have spent a lifetime learning and applying geological models and gathering data in the field and building digital datasets with the ultimate aim of a mineral discovery. It’s a fascinating and exciting topic as with the American tech companies investing heavily in the space, I also wonder if their studies rely on existing digital data common in the USA or are they also converting non digital data spread across the globe?

A few comments …..Firstly geologists are well acquainted with Big Data, we have been using terabyte datasets for probably a decade or more depending on the size-type of project etc. We have a range of expensive big softwares which are used to manipulate data in 3D, its not new.

Secondly, as with any data there is ‘great data’ ‘good data’ ‘not useful data’ and ‘incorrect data’ sorting through to usable digital format costs time and money, some data may even be difficult to define the big data ‘relationships’ needed for algorithm searches. Does running algorithms or dispersing data to crowdsourcing create a more precise and reliable exploration target when both rely on the same data inputs? Machine learning relies on ‘available geoscience’ however in the absence of data, can successful targets be defined?

Years ago at the beginning of the iron ore boom, I ran a workshop with 20-30 geologists each with variable experience and variable geological backgrounds to define iron ore targets in the Pilbara, Western Australia. We used regional datasets (hardcopy and GIS) and the participants were not familiar with the regional known deposits only the local ones they worked on. After spending two days going through regional geochemistry, geology, mineral occurrences, aerial-magnetics and pretty crappy ground gravity, the teams circled areas in which they believed could be mineralised for the various iron ore mineralisation styles (Marra Mamba, CID, Brockman & Nimingarra Fm) they were assigned. For most mineralisation styles the targets defined were the already existing mining areas and known deposits which the participants didn’t precisely know during the exercise. We were however looking for new deposits. What I learned during this exercise was that ‘existing data’ points geologists to familiarise themselves within areas where data exists and not where data doesn’t exist, there weren’t many new targets as there were large areas without data. Can crowdsourcing or algorithms validate exploration targets in the absence of data? Unlikely. The high cost of exploration is in the gathering of good data which is time consuming and costly. 4IR hasn’t reduced the high cost of exploration at this stage in its evolution, we’re still looking for ways to reduce drilling costs and have been trying to do for a decade with limited success.

Thirdly, deposits are located beneath the land surface where geologists struggle to see and even with modern technology the targets remain uncertain before drill testing. Can an algorithm or crowdsourcing do more or better seeing beneath the earth? Drilling and geophysics are likely the better tools, however again it comes down to the spread, depth and quality of data inputs. Exploration is essentially testing a hypothesis in the absence of data ie lets use various exploration tools to determine if and where to drill a hole and then drill test the idea to find a deposit, the decision on when and where to drill is not always an easy decision and most junior companies are in a hurry for success to keep shareholders content and the larger companies are generally not in a hurry as a new deposit doesn’t give much impact on their shareprice, its more about the tonnages and ounces they produce and annual dividends. It’s the indirect exploration activities and drilling which cost the big bucks and ultimately creates the risk in the exploration game as there is no guarantee for success. For exploration geologists, having the ability to see deposits beneath the surface for a range of minerals and commodities will certainly be a game changer and likely geophysics is closer than any other technique but it remains expensive, the larger or more detailed the survey, the more it costs.

In terms of geoscientific data, regional data creates regional targets, and project scale data creates project scale targets, both serve different targeting purposes. The deeper you explore and dig into the data pile ie from regional down to project scale drilling targets, the more information you have to run algorithms and provide for crowdsourcing ideas however the more expensive the exploration program. Those countries with well organised geoscientific data will definitely benefit however mineral endowment is driven by the earths crust processes and not the data systems used to manipulate. If you’re in a highly endowed terrain and you have great data, perhaps we’ll learn more about super terrains but those regions without good quality data will only get left behind and this isn’t positive for humanity in undeveloped countries.

The comparison to Linux where open source programming has provided large benefits for the program and Linux community, the difference with exploration data however is only the company that holds the mine tenure will benefit from a mineral discovery as it is funding the exploration expenditure in which to make the discovery, the invested interest is not equal weighted.

Ozminerals have certainly taken a big step forward, although AUD$1M is a lot of drill meters that could have been spent on testing existing targets for discovery and would also give a fabulous airborne or seismic survey or further high end data manipulation as well. Will they have the budget and persistence to test all 400 targets, unlikely, but they will learn and fine tune the process as they drill test the targets, if there is a great benefit they are well advanced.

Another aspect to consider is exploration licenses and permits are commonly granted for a specific time frame commonly two-three-five years with extensions, if a project is nearing the end of the permitted timeframe and have not succeeded in making a mineral discovery in-house, the project is more likely to enable public viewing of the data towards the end of the tenure period.

I believe there is a long way to go in terms of ability-enthusiasm to make exploration data public by mining companies due to the high cost of acquisition and likely a number of new breakthroughs to happen to ignite the enthusiasm. Time will tell, watch this space, its terribly interesting.

This article was prompted by a recent article by Stockhead, click on link below to view.

Stockhead article