Fresh Thinking: The 3-Minute Read. Episode 100

Busy day? Short on time?
Here’s your 3-minute quick read version of Fresh Thinking by Snowden Optiro’s podcast Episode 100.
We’ve pulled out the key insights and takeaways—perfect for when you want the value, minus the headphones.

Should We Minimise Conditional Bias in Resource Estimation?

Welcome to another Fresh Thinking podcast recap!

In this episode, Snowden Optiro’s Ian Glacken sat down with two of the sharpest minds in geostatistics and resource estimation—Graeme Lyall, Executive Consultant at Snowden Optiro, and Dr Oscar Rondon, Principal Geostatistician at Datamine.

They tackled one of the most debated—and often divisive—topics in mineral resource estimation:
Should we focus on minimising conditional bias, or would it be better to prioritise an accurate grade-tonnage relationship at the time of mining?

---

The Debate: What Is Conditional Bias?

Conditional bias refers to the varying difference between estimated grades in a block model and the true grades.
Unlike constant bias (such as uncalibrated assay equipment), conditional bias isn’t fixed—it changes depending on the grade or cut-off grade.

That variability often results in misclassification—sending ore to the waste dump and waste to the plant—leading to major downstream impacts.

---

Two Schools of Thought

Ian, Graeme, and Oscar broke down two opposing approaches:

1. Minimise Conditional Bias

This school of thought, led by pioneers like Krige and supported by Van et al. (2003), argues that conditional bias should be minimised across the board—whether it is a short-term or long-term model.
Tools like kriging neighbourhood analysis (KNA) and carefully tuned search parameters are key to keeping bias in check.

2. Accept It & Focus on Grade-Tonnage

On the other side, experts like Isaaks and Deutsch contend that conditional bias is inevitable.
Their view? Prioritise long-term grade-tonnage accuracy by using more restrictive searches and fewer samples—accepting some level of bias as a trade-off.

---

Smoothing, Block Sizes, and the Role of Reconciliation

The team also addressed common misconceptions—like the belief that more smoothing always equates to more conditional bias.

They explored:

• The relationship between block size selection and smoothing.
• How tools like the Discrete Gaussian Method are used to generate theoretical grade-tonnage curves.
• Why the ultimate validation always come down to reconciliation—comparing estimates with actual production data.

---

Key Takeaway

There’s no silver bullet.

The real skill lies in understanding the trade-offs—and being clear-eyed about the purpose of the model.

---

Further Reading

For those wanting to learn more, here are the key papers the team recommended:

• Krige (1996)
• Van et al. (2003)
• Isaaks (2005)
• Deutsch (2007)
• Noack & Luong (2017)

If you would like to contact Ian, Graeme or Oscar, please email them: contact@snowdenoptiro.com

---

Listen to the Full Episode

Curious to hear the full debate and insights?
You can watch/listen to the full episode of this Fresh Thinking by Snowden Optiro podcast on YouTube, Spotify, Apple Podcasts, Amazon Music and most other podcast platforms.