Resnikoff

In doing some reading about machine reliability, and risk-based predictive maintenance, I came across an interesting theory: the Resnikoff Conundrum. 

The theory comes from a mathematical study into reliability-based maintenance in 1978.  I think the theory also applies to quantitative risk management, exposing an issue with statistical risk management.

In order to collect failure data, there must be failures.  But serious failures are designed out, so, there’s no or very little data. Failures of critical items causing injury or death are unacceptable and are therefore designed out.

So: we are trying to prevent failure, based on data that requires failure.