🚀 go-pugleaf

Regarding the analogy between reliability of complex software and jets:
 
> like to point out that Boeing and other aircraft developers spend
>BILLIONS of dollars developgind (and performing SQA) on each new
>jet--jets which they then sell for HUNDREDS of MILLIONS dollars.  A major
>software package, on the other hand, costs less than 5 MILLION to develop
>and sells for HUNDREDS of dollars.  Yet, I suggest that the number of
>potential 'fatal' failure points in a major software package are probably
>greater than the number of potential fatal failure points in a modern jet.
 
It's exceedingly unlikely that there are fewer failure points in the jet, when
you consider the whole jet, particularly since the new ones have 10's to 100s
of computer systems in them.  What we don't often thing about is that lots of
the bits that go into it are standard/prefabricated/pre-tested/etc pieces that
don't have to be rigorously tested every time a new model is rolled out: tires,
brake assemblies, particular metallurgical formulations, control cables,
hydraulic assemblies, etc.  Perhaps the better analogy for a large software
system would be the space shuttle, which could only rely on a relatively _much_
smaller set of existing components (Remember the Challenger?).
 
I don't remember where this thread started, but it _has_ brought us around to
the need for _useable_ reusable software components.
 
Lee Harris