April 2019

The miner and his gold: a modern fable

  • By Holly Bridgwater MAusIMM, Industry Lead - Crowdsourcing, Unearthed

Is the industry missing out on unearthing buried treasure by hoarding their data?

In 1999, former CEO of Goldcorp Rob McEwen attended a lecture at Massachusetts Institute of Technology about the benefits of employing open source software (Afuah, 2009). Inspired by this framework and the value created by open collaboration, McEwen realised the significant potential for applying this concept in the mining industry.

Finding new exploration targets presents a key resources sector challenge that fits the open collaboration framework. In 1999, Goldcorp was a CAD$100 million gold mining company, with an under-producing mine, Red Lake. At the same time, the company was struggling to find new resources, despite a high-calibre exploration team. McEwen became aware that there might be more value locked in the exploration data than his geologists could extract. He released this data to the public, totalling a mere 400 MB back in 2000, with a CAD$575,000 incentive to review it and identify additional targets (HeroX, 2015). The result was thousands of submissions from geologists, mathematicians and physicists that identified an additional CAD$6 billion worth of gold on Goldcorp tenements (Idea Connection, 2009). The process reduced the potential exploration period by two to three years, but more importantly transformed Goldcorp from a struggling junior miner into an industry leader.

The lecture attended by McEwen back in 1999 was presented by the founder of the Linux operating system, Linus Torvalds. If you think you are not familiar with Linux – guess again. Twenty years on from that lecture, every Android phone runs on Linux, and if you have an iPhone instead, almost all webpages and applications operate on Linux. This software has achieved world domination – and it is open source. 

Goldcorp is probably a more familiar story. Following the competition, Goldcorp grew into a US$10 billion company, was purchased by Newmont in January 2019 – creating the largest global gold mining company at that time (Bloomberg, 2019) – and is nowadays considered one of the most innovative gold mining companies in the world. 

So, where does this leave mineral exploration? Apart from a few standout exceptions – including Integra Gold (HeroX, 2015), and now OZ Minerals (Unearthed Solutions, 2018) – it seems we have not yet learned the value of following in the footsteps of Linux and making the most of open collaboration. 

Finding hidden value

Like the old fable of the miser and his gold, value not used is value that does not exist. Many of us hold the belief that releasing our data publicly would give up our competitive advantage. It would allow other people to use our data to find other similar deposits to those that we are exploring for. But I would argue that we are not creating a true competitive advantage for ourselves by keeping our data private. The fact that in the past two decades there have been no major mineral discoveries in Australia (Mining Weekly, 2018) supports this position. 

Rather, it is possible that your model is missing something and that you have overlooked hidden value in your data.

If you own a tenement and you own the data, your competitive advantage should be your ability to make the best use of it; to most efficiently use that information to quickly and accurately define the resources within your land holding. There is a strong argument that publicly sharing data can create this advantage for you. 

Having external groups work on your data provides both qualitative and quantitative advantages. Qualitatively, people with different expertise, experience and knowledge develop models and interpretations in different ways. They provide novel approaches and insights into data that are more varied and broader than what we may develop through our own limited experience. An example of this is how a data scientist and geologist develop their models. While a data scientist will apply statistics with computer science and domain knowledge, geologists will combine knowledge and physics-based models, with statistical techniques. Both approaches are valid and provide different insights. 

Opening up data also gives a distinct numbers advantage. In the Integra GoldRush challenge, thousands of people were able to process 75 years of historical data and provide the best consensus model. This was done in months, rather than the years it would have taken the small internal Integra team.

Providing data to the public is not a standardised process

If you have come around to the idea that putting your data into the public domain can provide a lot more value, it’s worth exploring how you would go about doing that. Expanding organisational boundaries to welcome a plethora of talent to look at our data is not a process we are used to, nor one that fits within most companies’ exploration workflow. To date, we have seen companies approach this by running incentivised, competition-style projects: Goldcorp, Integra, and recently OZ Minerals. 

We need to ensure that the technology built to deliver these successful experiments is developed into a repeatable process that becomes more widely accessible and ingrained into the exploration workflow.

The key to making these open data projects scalable and successful is approachable data.

‘Quickly making data accessible to a community with diverse skills will create a serious shift in how exploration happens.’

Australia has one of the largest and most comprehensive publicly available mineral exploration databases in the world. In most states, when an area of land is relinquished, the data that was collected during the period of tenure is released to the public. 

Therefore, we have a huge resource at our fingertips to assess new projects and plan our first phases of exploration. Yet, as any geologist that has done this work will attest to, this can take a very long time. The work involves searching through old reports and pulling together all the drilling results from different companies, with different analytical techniques at different detection limits. To process historical data and use it meaningfully can take months or years. 

So, although there might be a lot of data, if we did not collect it ourselves, it could be highly inaccessible. 

Some exploration teams become so exasperated by this process that they simply go and spend money to collect the same data afresh. It is not uncommon for a new geologist to uncover a key piece of historical data three years into an exploration project.

So, when we put our data to the public – many of whom are not geoscience domain experts – that data needs to be accessible, searchable and have context for people to be able to use it. Data that fits these criteria can provide the scalable platform to consistently provide exploration data in a meaningful way to the public. We can then allow people with different skill sets, such as data science and machine learning, to apply their models to our data and gain value from their interpretations.

There are obvious benefits from more accessible exploration data: 

  • easier access for data scientists to apply more and variable models to interpret the data
  • faster assessment of pre-competitive data for mergers and acquisitions
  • the ability for regulators to showcase the prospectivity of the ground within their state for certain minerals.

Powerful tools for quickly making data accessible to a community with diverse skills will create a serious shift in how exploration happens.

The path forward

We are seeing a significant upward trend in the applications of machine learning to the areas of exploration targeting and deposit prediction. The concepts of big data and open data are therefore becoming increasingly important. Machine learning algorithms are only ever as good as the data they are provided. The more data points they are able to learn from, the more accurate their predictions become. If you think you can build a machine learning model from the signature of one small deposit to predict where the next one is, forget it. The success of emerging machine learning exploration businesses such as OreFox, Earth AI and Solve Geosolutions is down to their ability to use large data sources, and these typically are not only private, but also public. 

Therefore, the most effective machine learning models for targeting will be those trained on the largest and most varied datasets. If that is the case, publicly opening your data could provide a competitive advantage for those developing machine learning models, by making their models more accurate. It’s true that there is a risk of allowing other people to develop models that predict the location of deposits you want to find, but it is a challenge to find any organisation that has avoided disruption from technology by doing nothing. Do you really trust everyone else in the market to do nothing as well? It is already clear that is not the case, but technology dissemination does not happen overnight.

As we see this shift towards using machine learning, modern companies such as OZ Minerals, Integra and Goldcorp are leading the market and taking advantage of skills and new approaches by making their data publicly available, as well as employing a variety of data science approaches internally. As mentioned above, we are also seeing new types of companies enter the market, merging machine learning and exploration. Additionally, regulators are realising the potential of these technologies to demonstrate the prospectivity of their regions and pushing for more open data. 

There is little doubt that as an industry, open data would significantly increase our discovery rate. More data equals better and more accurate models. But if there is a significant industry shift like this, and everyone benefits for a short time, in the long run it becomes difficult for an organisation to maintain a competitive advantage. We may see a fundamental change in the exploration business model. 

Things may be changing faster than we think for mineral exploration. How is your company going to ensure it can capitalise on this change? Are you missing out on billions of dollars of value by hoarding your data? 

References

Afuah A, 2009. Strategic Innovation: New Game Strategies for Competitive Advantage. 1st ed. Routledge.

Bloomberg 2019, Newmont to Buy Goldcorp in $10 Billion Deal, Creating World’s Largest Gold Miner [online]. Available from: https://www.bloomberg.com/news/articles/2019-01-14/newmont-to-buy-goldcorp-in-deal-valued-at-10-billion-jqw6a4yy [Accessed Jan 15 2019]

HeroX, 2015. Integra Gold Rush [online]. Available from: https://www.herox.com/IntegraGoldRush/ [Accessed Jan 15 2019]

Idea Connection, 2009, Open Innovation: Goldcorp Challenge [ONLINE] Available at: https://www.ideaconnection.com/open-innovation-success/Open-Innovation-Goldcorp-Challenge-00031.html [Accessed Jan 15 2019]

Mining Weekly, 2018. Australia seeks to reverse mineral discovery drought [Online] Available at: http://m.miningweekly.com/article/australia-seeks-to-reverse-mineral-discovery-drought-2018-06-13/rep_id:3861 [Accessed Jan 15 2019]

Unearthed Solutions, 2018. Explorer Challenge [Online]. Available from: https://unearthed.solutions/u/explorer-challenge [Accessed Jan 15 2019]

 

About the author

Holly Bridgwater is an exploration geologist, crowdsourcing lead at Unearthed, and an advocate for industry adoption of open data initiatives. Holly is working with OZ Minerals to deliver the Explorer Challenge, an online competition that invites innovators to test the limits of geology and data science to find new ways to explore. For more information, visit: https://unearthed.solutions/u/competitions/explorer-challenge. 

Share This Article