Writing Classes - Java Software Solutions: Foundations of Program Design - page 206

What Caused It?
The fiasco this project represents is considered to be one of the greatest software
engineering and overall system-design failures in history. There were many indi-
vidual issues that contributed to the problems, but in general, the designers simply
had too much faith in the technology they were using. They didn't factor in errors
and inefficiencies that always occur in a complex system.
The individual problems included the fact that the software misinterpreted data
from photoelectric eyes and, therefore, did not detect a pile of existing bags. When
the system was restarted after it crashed, it lost information about the status of
carts and didn't know which ones were full or not. Sharp corners were not fac-
tored in correctly regarding the speed of the conveyors, resulting in spilled carts.
A telescoping belt loader called the lizard tongue—designed to reach into a planes
cargo hold and pick up bags without human assistance—failed completely.
Lessons Learned
A project of this scope should not be attempted without many localized tests
on the technologies in use. The failures were often a result of multiple variables
caused by the system as a whole in operation.
When the system was designed in the late 1980s, it relied on a centralized main-
frame to control the system. Such an approach seems ridiculous in today's world
of distributed processing, but it is a good example of how a large system can be
obsolete by the time it is developed. Today, human baggage handlers with hand-
held scanners result in a far more fast and accurate delivery system than the best
goals of the planned automated system.
Source: The International Herald Tribune