Originally Posted by
Bushfiva
Is there any truth in Private Eye's observations on p29 of issue 1204? Is it normal to have redundant software written by different companies in the commercial aerospace industry?
On the first, I don't know. Porus chinese walls I could believe, ditching triplex development completely I would find slightly suprising. I don't know the details though.
As to the second question - yes. Although it may be better stated that normal would be separate teams (which may be in same company). Typically the teams would be in different sites / locations etc. (regardless of company). The geographical separation probably doesn't buy you as much these days.
The technique is known as multi-version or N-version development. The assumptions (eg. that separate development reduces common-mode failure risk) that underly it have been questioned - look up Knight Leveson experiment.
All of which, though interesting, is of
no relevance to BA38, as the AAIB have already established that flight data shows all the software working correctly right through to opening of the fuel valves. The only possibility for software failure now is if it was a type of failure that caused the engines not to respond as commanded
and simultaneously caused
all systems to send fake data to QAR and FDR that
everything was normal. Simultaneous (or nearly) flying saucer ingestion is more probable.