[SystemSafety] Fault, Failure and Reliability Again (short)

[Peter Bernard Ladkin]

Folks,

a partial reply to Nick is at
http://www.abnormaldistribution.org/2015/03/04/fault-failure-reliability-definitions/

John Knight said he uses the Laprie taxonomy (as the 2004 paper referenced by Örjan yesterday is
known) in his book, precisely in order to keep straight on such matters as we have been discussing.
I took John's suggestion in my blog post. Software failure, software fault, software error, and
software reliability are given a clear meaning in the Laprie taxonomy, although it is not determined
how one may measure software reliability.

Since the Laprie taxonomy is the result of decades of deliberate in IFIP WG 10.4 which first
resulted in a book, before this paper, it should by no means be dismissed lightly.

This decisively contradicts Nick's proposal that

On 2015-03-03 22:50 , Nick Tudor wrote:
> .....the software does not fail, the system does.  It therefore makes no sense to talk about
> reliability of software 

Nick has suggested privately that this view is prevalent in aerospace, and I am informed indirectly
that it is also prevalent in some areas of the nuclear industry. If so, it is surely a matter of
engineering reasonsibility due diligence that these engineers get themselves up to date on the
concepts the experts, the computer scientists, use to talk about their subject matter.

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Je suis Charlie
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de