How are software packages developed? - The Tao of Computing - page 139

4. What would happen if you knew that the software upon
which you were currently depending malfunctioned 5% of
the time? Would you behave differently?
5. Consider each of the following reports regarding early ver
sions of the F16 fighter plane. In each case, speculate about
what stage of the software development cycle (specifications,
algorithm design, coding) might have generated the difficulty,
and explain your answer.
a. The F16 fighter plane uses a computer to help control the rais
ing and lowering of its landing gear, and the hardware and
software have worked effectively in accomplishing this task. In
an early version, however, the assumption was made that the
plane would always be flying whenever the request was issued
to raise the wheels. Unfortunately, violation of this unwritten
assumption early in the testing of the plane caused significant
damage. (Since that time, a new procedure, weightonwheels
has been added to check this assumption.)*
b. Control of the F16 is under the direction of a computer so
that the technologically advanced machine will be able to re
spond immediately to situations it encounters while flying.
In an early version of the plane, however, the navigation sys
tem did not distinguish between the northern hemisphere
and the southern hemisphere, and the plane would have
flipped over when it crossed the equator.**
6. Suppose you rely on a software package to do various com
putations as part of your work. This question encourages dis
cussion regarding legal liabilities involving software updates.
a. If you know your current software sometimes generates an
error and you learn that this problem is fixed in a new ver
sion, do you think you should be legally liable for damages
if you do not upgrade your software as soon as possible?
b. Suppose it seems your current software handles process
ing correctly. A new version of the software comes out,
*From The Limits of Computing , pp. 17-18. Other problems with the
F16 are reported in Software Engineering Notes , October 1986, p. 6.
**From The Limits of Computing , p. 20. This story was reported in a
letter by Earl Boebert in Software Engineering Notes , Volume 6,
Number 2, April 1980, p. 5.