[SystemSafety] Fetzer

[andy at the-ashworths.org]

Derek,

> 
> Martyn,
> 
> > Why is it that people (standards committees, for example) are happy to
> > support unscientific claims based on testing and process conformance
> > but reject the need for scientific claims based on mathematical proof?
> 
> Mathematical proofs contain mistakes, because people make mistakes when
> writing then, just like writing code.

Test specifications, results analysis, and process conformance plans and
audit reports contain mistakes, because people make mistakes when writing
then (sic), just like writing code. 

> 
> Mistakes may be made in mapping the world into a mathematical model, i.e.,
> important aspects of reality are abstracted away or ignored.

Mistakes may be made in interpreting the significance of test results and
process conformance reports, i.e., important aspects of reality are
discounted or ignored.
> 
> Mistakes are made in writing the proof.
Mistakes are made in writing test / process conformance reports.

> 
> Mistakes are made in proving the proof.
> 
> Mistakes are made in interpreting what a proof means.

Mistakes are made in interpreting test / process conformance reports.

To me it seems that you are arguing that it is ok to make mistakes in
subjective areas such as testing and process conformance, yet formal proof
must be held to a higher standard. 

There is a place for both proof and the more subjective analyses such as
testing and process conformance.

Cheers

Andy