[SystemSafety] Feynman and the Challenger disaster

[Andy Ashworth]

Another point raised by Feynman in his observations is that 1 in 100000 is equivalent to launching a shuttle every day for nearly 300 years and experiencing a single loss. Many people in safety talk of "10^-9" failure rates without putting this into the context of their own application. All too often, in my recent experience, engineers want a single number they can cite (usually out of context!) that clears them of having to make decisions or take responsibility. 

I have colleagues who want certification for everything and others who make arbitrary assignments of SIL and pass these onto our suppliers. The understanding of quantified risk today appears to be poor and I would suggest industry is making products that pose a higher risk than perhaps those of 25 years ago when system safety management was in its infancy. Where did we go wrong?

Andy Ashworth



Sent from Andy's iPad

> On Jun 15, 2015, at 16:12, Gergely Buday <gbuday at gmail.com> wrote:
> 
> Hi,
> 
> I came across this today:
> 
> "A fascinating 1986 text by Richard Feynman on the question of how to
> estimate the reliability of a space shuttle. This text was written as
> part of an investigation of the Challenger launch disaster. It
> describes a large mismatch between the reliability estimates made by
> the engineers (of the order of 1 failure in 100 uses) and the
> reliability estimates communicated to the public by the NASA
> management (of the order of 1 failure in 100 000 uses). I think this
> text should be of interest to anyone interested in security,
> engineering, the production of safety-critical products and its
> management."
> 
> https://plus.google.com/115261498097261634859/posts/RhApPYEDhvk
> 
> http://science.ksc.nasa.gov/shuttle/missions/51-l/docs/rogers-commission/Appendix-F.txt
> 
> - Gergely
> _______________________________________________
> The System Safety Mailing List
> systemsafety at TechFak.Uni-Bielefeld.DE
>