Saturday, September 29, 2012

Setting new standards?

The Bre-X fraud inspired the Toronto Stock Exchange (TSE) and the Ontario Securities Commission (OSC) to set up a task force. Its objective was to take a close look at National Instrument 43-101. The Members of the task force are given in this Interim Report. Mr Morley P Carscallen, OSC’s Vice Chair, and Mr John W Carson, TSE’s Senior Vice-President, took on this task in April 1997. It is a fact that Bre-X’s bogus gold grades and Busang’s barren rock were made to look by hook and by crook like a gold resource. But who were the crooks? And who set the hook for Bre-X’s shareholders? OSC’s own qualified persons have yet to grasp the fact that geostatistics is a scientific fraud! Perhaps ironically, it was geostatistical software that made Bre-X’s bogus grades and Busang’s barren rock to look like massive gold resource!

I have put on paper why geostatistics is a scientific fraud. A few simple steps were all it took to cook it up! The first step was to strip the variance off the distance-weighted average. The second step was to call what was left a kriged estimate to honor D G Krige and his work. Matheron taught his disciples how to work with infinite sets of kriged estimates and zero kriging variances. What a shame that such a simple scientific fraud underpinned what was called a new science. Matheron himself never got into counting degrees of freedom. Neither did Stanford’s Journel, UBC’s Sinclair, and similarly gifted scholars.  

Young Dr A J Sinclair took to geostatistics in the 1970s. He may well have thought that Matheron had a fresh take on applied statistics. In those days Sinclair was entrusted with teaching UBC’s students all about Earth Sciences. CIM Bulletin asked Sinclair in 1990 to review Precision Estimates for Ore Reserves. My son and I had shown how to test for spatial dependence between a set of gold grades determined in ordered rounds in a drift. Given that interleaved bulk samples had not been selected, it was impossible to estimate the intrinsic variance of gold. Professor Dr A J Sinclair, PEng, PGeo rejected our article. We were pleased that it was praised by and published in Erzmetall, October 1991.

What a surprise that David’s peers wanted to praise his 1977 Geostatistical Ore Reserve Estimation! Why would his peers want to praise infinite sets of simulated values? The stage for an international forum was set at McGill University on June 3-5, 1993. It was called Geostatistics for the Next Century. What is so striking in retrospect is the fact that Bre-X Minerals was already drilling in Borneo when David was praised by his peers! Nobody was interested in the properties of variances in 1993! Yet, the additive property of variances in a measurement chain played a key role in unscrambling the Bre-X fraud.

 Measurement variance included

 Measurement variance subtracted

The Mining Standard Task Force released its Final Report in January 1999. Why had MSTF not pointed out that geostatistical software had convert Bre-X’s bogus grades and Busang’s barren rock so slickly into a massive phantom gold resource? MSTF’s Final Report was made public in January 1999. On a positive note, Dr A J Sinclair no longer graces National Instrument 43-101. On a negative note, Setting New Standards still didn’t explain at all how the Bre-X fraud could have been nipped in the bud. So it was that the Mining Standard Task Force ended up as a farce. The properties of variances were nowhere to be found. Sinclair still teaches students at UBC's Department of Earth and Ocean Sciences how to assume spatial dependence, krige, smooth, and rig the rules of applied statistics with impunity. So much for scientific integrity!

I have set up several sources of information on my website. Under Correspondence are listed all sorts of letters in a context of source and time. Academic freedom to teach a scientific fraud makes no sense at all. The fact that "geostatistics has flourished in the scientific literature for more than four decades" does not imply that spatial dependence between measured values in ordered sets may be assumed. Neither does it imply that degrees of freedom need not be counted.

No comments: